In [ ]:
plot_confusion_matrix(Random_Forest_classifier, X_test1, y_test1,cmap="YlGnBu");
!git clone https://github.com/salohnana2018/Data-Science-Campus-Capstone-Project.git
!pip install boruta
#!pip install tsai
!pip install ta
stable = True # Set to True for latest pip version or False for main branch in GitHub
!pip install {"tsai -U" if stable else "git+https://github.com/timeseriesAI/tsai.git"}
import pandas as pd
from ta.momentum import RSIIndicator,StochasticOscillator
import numpy as np
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import seaborn as sns
from datetime import datetime
import warnings
warnings.filterwarnings('ignore')
import plotly.graph_objects as go
from plotly.subplots import make_subplots
from plotly.offline import download_plotlyjs, init_notebook_mode, plot, iplot
## to add new features
from ta.momentum import RSIIndicator,StochasticOscillator
# ML Part
import xgboost as xgb
from xgboost import XGBClassifier
from xgboost import XGBRegressor
from sklearn.model_selection import RandomizedSearchCV
from collections import Counter
from sklearn.svm import SVC
from sklearn.model_selection import train_test_split
from sklearn.preprocessing import StandardScaler
from sklearn.linear_model import LogisticRegression
from sklearn.metrics import plot_confusion_matrix
from sklearn.metrics import accuracy_score
from sklearn.metrics import confusion_matrix
from sklearn.linear_model import LinearRegression
from sklearn.metrics import mean_absolute_error
from sklearn.metrics import mean_squared_error
from sklearn.metrics import log_loss
from sklearn.model_selection import cross_val_score
from sklearn.metrics import r2_score
from sklearn.metrics import f1_score
from sklearn.metrics import classification_report
from sklearn.metrics import accuracy_score
from sklearn.preprocessing import LabelEncoder
from sklearn.pipeline import Pipeline
from sklearn.pipeline import make_pipeline
from sklearn.compose import ColumnTransformer
# Search CV
from sklearn.model_selection import GridSearchCV
from sklearn.model_selection import RandomizedSearchCV
from sklearn.linear_model import RidgeClassifierCV
from sklearn.dummy import DummyClassifier
from sklearn.model_selection import KFold
from boruta import BorutaPy
from scipy import stats
from sklearn.metrics import make_scorer, roc_auc_score
from sklearn.ensemble import RandomForestClassifier
#time series classification
from tsai.all import *
computer_setup()
print('✔️ Libraries Loaded!')
os : Linux-5.4.188+-x86_64-with-Ubuntu-18.04-bionic python : 3.7.13 tsai : 0.3.1 fastai : 2.7.10 fastcore : 1.5.22 torch : 1.11.0+cu102 device : cpu cpu cores : 2 RAM : 12.68 GB GPU memory : N/A ✔️ Libraries Loaded!
df_cleaned = pd.read_csv('Data-Science-Campus-Capstone-Project/Data2/cleaned_data_frame.csv')
df_Aramco=df_cleaned[df_cleaned['Name']=='Saudi Aramco']
df_SARCO=df_cleaned[df_cleaned['Name']=='SARCO']
df_Rabigh=df_cleaned[df_cleaned['Name']=='Petro Rabigh']
df_Bahri=df_cleaned[df_cleaned['Name']=='Bahri']
df_Aldrees=df_cleaned[df_cleaned['Name']=='Aldrees']
# function to split data.
def split_data(df, target):
# split the dataset into Training and Testing sets
# target is the name of the target column ==> 'Target'
X_train, X_test, y_train, y_test = train_test_split(
df.drop(['Tomorrow','Date','Symbol','Name','Change %','Open','Close','High','Low',target], axis=1), # drop the target
df[target], # choosing our target
train_size = 0.8,
test_size=0.2,
#stratify=df[target],
random_state=42)
return X_train, X_test, y_train, y_test
# scale numeric data.
def scale_data(X_train, X_test):
num_features = X_train.describe(include=[np.number]).columns
scaler = StandardScaler()
X_train = scaler.fit_transform(X_train[num_features])
X_test = scaler.transform(X_test[num_features])
return X_train, X_test
X_train1, X_test1, y_train1, y_test1 = split_data(df_cleaned, 'Target')
X_train1_sc, X_test1_sc = scale_data(X_train1, X_test1)
# Number of trees in random forest
n_estimators = [200]
# Number of features to consider at every split
max_features = ['auto', 'sqrt', None, 'log2']
# Maximum number of levels in tree
max_depth = [40]
max_depth.append(None)
# Minimum number of samples required to split a node
min_samples_split = [1, 10, 20]
# Minimum number of samples required at each leaf node
min_samples_leaf = [1, 14, 16 ,20]
# Method of selecting samples for training each tree
bootstrap = [True, False]
# Create the random grid
random_grid = {'n_estimators': n_estimators,
'max_features': max_features,
'max_depth': max_depth,
'min_samples_split': min_samples_split,
'min_samples_leaf': min_samples_leaf,
'bootstrap': bootstrap}
print(random_grid)
{'n_estimators': [200], 'max_features': ['auto', 'sqrt', None, 'log2'], 'max_depth': [40, None], 'min_samples_split': [1, 10, 20], 'min_samples_leaf': [1, 14, 16, 20], 'bootstrap': [True, False]}
# New Random Forest Classifier to house optimal parameters
forest = RandomForestClassifier(class_weight='balanced')
# Specfiy the details of our Randomized Search
forest = RandomizedSearchCV(estimator = forest, param_distributions = random_grid, n_iter = 100, cv = 10, verbose=2, random_state=42, n_jobs = -1)
# Fit the random search model
forest.fit(X_train1_sc, y_train1)
Fitting 10 folds for each of 100 candidates, totalling 1000 fits
RandomizedSearchCV(cv=10,
estimator=RandomForestClassifier(class_weight='balanced'),
n_iter=100, n_jobs=-1,
param_distributions={'bootstrap': [True, False],
'max_depth': [40, None],
'max_features': ['auto', 'sqrt', None,
'log2'],
'min_samples_leaf': [1, 14, 16, 20],
'min_samples_split': [1, 10, 20],
'n_estimators': [200]},
random_state=42, verbose=2)In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. RandomizedSearchCV(cv=10,
estimator=RandomForestClassifier(class_weight='balanced'),
n_iter=100, n_jobs=-1,
param_distributions={'bootstrap': [True, False],
'max_depth': [40, None],
'max_features': ['auto', 'sqrt', None,
'log2'],
'min_samples_leaf': [1, 14, 16, 20],
'min_samples_split': [1, 10, 20],
'n_estimators': [200]},
random_state=42, verbose=2)RandomForestClassifier(class_weight='balanced')
RandomForestClassifier(class_weight='balanced')
forest.best_params_
{'n_estimators': 200,
'min_samples_split': 20,
'min_samples_leaf': 20,
'max_features': None,
'max_depth': 40,
'bootstrap': True}
###initialize Boruta
# the hyper-bramters was selected using grid search
forest = RandomForestClassifier(
n_jobs = -1,
n_estimators = 200,
min_samples_split= 20,
min_samples_leaf=14,
max_features='log2',
max_depth = 40,
bootstrap = False
)
boruta = BorutaPy(
estimator = forest,
n_estimators = 200,
max_iter = 1000 # number of trials to perform
)
### fit Boruta (it accepts np.array, not pd.DataFrame)
boruta.fit(np.array(X_train1_sc), np.array(y_train1))
### print results
BorutaPy(estimator=RandomForestClassifier(bootstrap=False, max_depth=40,
max_features='log2',
min_samples_leaf=14,
min_samples_split=20,
n_estimators=200, n_jobs=-1,
random_state=RandomState(MT19937) at 0x1D69E64B340),
max_iter=1000, n_estimators=200,
random_state=RandomState(MT19937) at 0x1D69E64B340)In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. BorutaPy(estimator=RandomForestClassifier(bootstrap=False, max_depth=40,
max_features='log2',
min_samples_leaf=14,
min_samples_split=20,
n_estimators=200, n_jobs=-1,
random_state=RandomState(MT19937) at 0x1D69E64B340),
max_iter=1000, n_estimators=200,
random_state=RandomState(MT19937) at 0x1D69E64B340)RandomForestClassifier(bootstrap=False, max_depth=40, max_features='log2',
min_samples_leaf=14, min_samples_split=20,
n_estimators=200, n_jobs=-1,
random_state=RandomState(MT19937) at 0x1D69E64B340)RandomForestClassifier(bootstrap=False, max_depth=40, max_features='log2',
min_samples_leaf=14, min_samples_split=20,
n_estimators=200, n_jobs=-1,
random_state=RandomState(MT19937) at 0x1D69E64B340)strong_features = X_train1.columns[boruta.support_].to_list()
cant_decide_features = X_train1.columns[boruta.support_weak_].to_list()
print('strongst_features:', strong_features)
print('features that boruta can not decide if its storng or weak :', cant_decide_features)
strongst_features: ['Volume', 'EMA_9', 'SMA_5', 'SMA_10', 'SMA_15', 'SMA_20', 'SMA_25', 'SMA_30', 'MACD', 'RSI', 'Stochastic', 'Open-Close', 'High-Low'] features that boruta can not decide if its storng or weak : []
X_train_base, X_test_base, y_train_base, y_test_base = split_data(df_Aramco, 'Target')
X_train_sc,X_test_sc = scale_data(X_train_base, X_test_base)
dummy_clf = DummyClassifier(strategy="prior",random_state=42)
dummy_clf.fit(X_train_sc, y_train_base)
preds = dummy_clf.predict(X_test_base)
score1 = dummy_clf.score(X_test_sc, y_test_base)*100
print("Baseline Score: {}".format(score1))
print(classification_report(y_test_base,preds))
Baseline Score: 59.70149253731343
precision recall f1-score support
0 0.00 0.00 0.00 54
1 0.60 1.00 0.75 80
accuracy 0.60 134
macro avg 0.30 0.50 0.37 134
weighted avg 0.36 0.60 0.45 134
plot_confusion_matrix(dummy_clf, X_test_base, y_test_base,cmap="YlGnBu");
dummy_clf = DummyClassifier(strategy="uniform",random_state=42)
dummy_clf.fit(X_train_sc, y_train_base)
preds = dummy_clf.predict(X_test_base)
score = dummy_clf.score(X_test_sc, y_test_base)
print("Baseline Score: {}".format(score))
print(classification_report(y_test_base,preds))
Baseline Score: 0.44029850746268656
precision recall f1-score support
0 0.32 0.33 0.32 54
1 0.53 0.51 0.52 80
accuracy 0.44 134
macro avg 0.42 0.42 0.42 134
weighted avg 0.45 0.44 0.44 134
plot_confusion_matrix(dummy_clf, X_test_base, y_test_base,cmap="YlGnBu");
X_train_base, X_test_base, y_train_base, y_test_base = split_data(df_SARCO, 'Target')
X_train_sc,X_test_sc = scale_data(X_train_base, X_test_base)
dummy_clf = DummyClassifier(strategy="prior",random_state=42)
dummy_clf.fit(X_train_sc, y_train_base)
preds = dummy_clf.predict(X_test_sc)
score2 = dummy_clf.score(X_test_sc, y_test_base)*100
print("Baseline Score: {}".format(score2))
print(classification_report(y_test_base,preds))
Baseline Score: 94.16498993963782
precision recall f1-score support
0 0.94 1.00 0.97 468
1 0.00 0.00 0.00 29
accuracy 0.94 497
macro avg 0.47 0.50 0.48 497
weighted avg 0.89 0.94 0.91 497
plot_confusion_matrix(dummy_clf, X_test_base, y_test_base,cmap="YlGnBu");
dummy_clf = DummyClassifier(strategy="uniform",random_state=42)
dummy_clf.fit(X_train_sc, y_train_base)
preds = dummy_clf.predict(X_test_base)
score = dummy_clf.score(X_test_sc, y_test_base)
print("Baseline Score: {}".format(score))
print(classification_report(y_test_base,preds))
Baseline Score: 0.4869215291750503
precision recall f1-score support
0 0.94 0.49 0.64 468
1 0.06 0.48 0.10 29
accuracy 0.49 497
macro avg 0.50 0.48 0.37 497
weighted avg 0.89 0.49 0.61 497
plot_confusion_matrix(dummy_clf, X_test_base, y_test_base,cmap="YlGnBu");
X_train_base, X_test_base, y_train_base, y_test_base = split_data(df_Rabigh, 'Target')
X_train_sc,X_test_sc = scale_data(X_train_base, X_test_base)
dummy_clf = DummyClassifier(strategy="prior",random_state=42)
dummy_clf.fit(X_train_sc, y_train_base)
preds = dummy_clf.predict(X_test_sc)
score3 = dummy_clf.score(X_test_sc, y_test_base)*100
print("Baseline Score: {}".format(score3))
print(classification_report(y_test_base,preds))
Baseline Score: 96.579476861167
precision recall f1-score support
0 0.00 0.00 0.00 17
1 0.97 1.00 0.98 480
accuracy 0.97 497
macro avg 0.48 0.50 0.49 497
weighted avg 0.93 0.97 0.95 497
plot_confusion_matrix(dummy_clf, X_test_base, y_test_base,cmap="YlGnBu");
dummy_clf = DummyClassifier(strategy="uniform",random_state=42)
dummy_clf.fit(X_train_sc, y_train_base)
preds = dummy_clf.predict(X_test_base)
score = dummy_clf.score(X_test_sc, y_test_base)
print("Baseline Score: {}".format(score))
print(classification_report(y_test_base,preds))
Baseline Score: 0.5090543259557344
precision recall f1-score support
0 0.03 0.47 0.06 17
1 0.96 0.51 0.67 480
accuracy 0.51 497
macro avg 0.50 0.49 0.36 497
weighted avg 0.93 0.51 0.65 497
plot_confusion_matrix(dummy_clf, X_test_base, y_test_base,cmap="YlGnBu");
X_train_base, X_test_base, y_train_base, y_test_base = split_data(df_Bahri, 'Target')
X_train_sc,X_test_sc = scale_data(X_train_base, X_test_base)
dummy_clf = DummyClassifier(strategy="prior",random_state=42)
dummy_clf.fit(X_train_sc, y_train_base)
preds = dummy_clf.predict(X_test_sc)
score4 = dummy_clf.score(X_test_sc, y_test_base)*100
print("Baseline Score: {}".format(score4))
print(classification_report(y_test_base,preds))
Baseline Score: 55.33199195171026
precision recall f1-score support
0 0.00 0.00 0.00 222
1 0.55 1.00 0.71 275
accuracy 0.55 497
macro avg 0.28 0.50 0.36 497
weighted avg 0.31 0.55 0.39 497
plot_confusion_matrix(dummy_clf, X_test_base, y_test_base,cmap="YlGnBu");
dummy_clf = DummyClassifier(strategy="uniform",random_state=42)
dummy_clf.fit(X_train_sc, y_train_base)
preds = dummy_clf.predict(X_test_base)
score = dummy_clf.score(X_test_sc, y_test_base)
print("Baseline Score: {}".format(score))
print(classification_report(y_test_base,preds))
Baseline Score: 0.5150905432595574
precision recall f1-score support
0 0.46 0.50 0.48 222
1 0.57 0.52 0.54 275
accuracy 0.52 497
macro avg 0.51 0.51 0.51 497
weighted avg 0.52 0.52 0.52 497
plot_confusion_matrix(dummy_clf, X_test_base, y_test_base,cmap="YlGnBu");
X_train_base, X_test_base, y_train_base, y_test_base = split_data(df_Aldrees, 'Target')
X_train_sc,X_test_sc = scale_data(X_train_base, X_test_base)
dummy_clf = DummyClassifier(strategy="prior",random_state=42)
dummy_clf.fit(X_train_sc, y_train_base)
preds = dummy_clf.predict(X_test_sc)
score5 = dummy_clf.score(X_test_sc, y_test_base)*100
print("Baseline Score: {}".format(score5))
print(classification_report(y_test_base,preds))
Baseline Score: 63.983903420523134
precision recall f1-score support
0 0.64 1.00 0.78 318
1 0.00 0.00 0.00 179
accuracy 0.64 497
macro avg 0.32 0.50 0.39 497
weighted avg 0.41 0.64 0.50 497
plot_confusion_matrix(dummy_clf, X_test_base, y_test_base,cmap="YlGnBu");
dummy_clf = DummyClassifier(strategy="uniform",random_state=42)
dummy_clf.fit(X_train_sc, y_train_base)
preds = dummy_clf.predict(X_test_base)
score = dummy_clf.score(X_test_sc, y_test_base)
print("Baseline Score: {}".format(score))
print(classification_report(y_test_base,preds))
Baseline Score: 0.5231388329979879
precision recall f1-score support
0 0.67 0.51 0.58 318
1 0.39 0.55 0.45 179
accuracy 0.52 497
macro avg 0.53 0.53 0.52 497
weighted avg 0.57 0.52 0.53 497
plot_confusion_matrix(dummy_clf, X_test_base, y_test_base,cmap="YlGnBu");
# split data
X_train1, X_test1, y_train1, y_test1 = split_data(df_Aramco, 'Target')
#scaling the data
X_train1, X_test1 = scale_data(X_train1, X_test1)
# split data
X_train2, X_test2, y_train2, y_test2 = split_data(df_SARCO, 'Target')
#scaling the data
X_train2, X_test2 = scale_data(X_train2, X_test2)
# split data
X_train3, X_test3, y_train3, y_test3 = split_data(df_Rabigh, 'Target')
#scaling the data
X_train3, X_test3 = scale_data(X_train3, X_test3)
X_train4, X_test4, y_train4, y_test4 = split_data(df_Bahri, 'Target')
#scaling the data
X_train4, X_test4 = scale_data(X_train4, X_test4)
X_train5, X_test5, y_train5, y_test5 = split_data(df_Aldrees, 'Target')
#scaling the data
X_train5, X_test5 = scale_data(X_train5, X_test5)
Random_Forest_classifier = RandomForestClassifier(n_estimators=300, min_samples_split=100, criterion = 'gini', random_state=42,class_weight="balanced")
Random_Forest_classifier.fit(X_train1, y_train1)
RandomForestClassifier(class_weight='balanced', min_samples_split=100,
n_estimators=300, random_state=42)In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. RandomForestClassifier(class_weight='balanced', min_samples_split=100,
n_estimators=300, random_state=42)# training set
y_pred_RF_tr1 = Random_Forest_classifier.predict(X_train1)
accuracy_RF_tr1 = accuracy_score(y_train1, y_pred_RF_tr1)
print("Random Forest Classification Model Score (Training Set): %.2f%%" % (accuracy_RF_tr1 * 100.0))
# testing set
y_pred_RF_t1 = Random_Forest_classifier.predict(X_test1)
predictions_RF1 = [round(value) for value in y_pred_RF_t1]
# evaluate predictions
accuracy_RF_t1 = accuracy_score(y_test1, predictions_RF1)
print("Random Forest Classification Model Score (Testing Set): %.2f%%" % (accuracy_RF_t1 * 100.0))
Random Forest Classification Model Score (Training Set): 72.57% Random Forest Classification Model Score (Testing Set): 64.93%
plot_confusion_matrix(Random_Forest_classifier, X_test1, y_test1,cmap="YlGnBu");
# Printing the classification report
print(classification_report(y_test1, y_pred_RF_t1))
precision recall f1-score support
0 0.56 0.57 0.57 54
1 0.71 0.70 0.70 80
accuracy 0.65 134
macro avg 0.64 0.64 0.64 134
weighted avg 0.65 0.65 0.65 134
Random_Forest_classifier.fit(X_train2, y_train2)
RandomForestClassifier(class_weight='balanced', min_samples_split=100,
n_estimators=300, random_state=42)In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. RandomForestClassifier(class_weight='balanced', min_samples_split=100,
n_estimators=300, random_state=42)# training set
y_pred_RF_tr2 = Random_Forest_classifier.predict(X_train2)
accuracy_RF_tr2 = accuracy_score(y_train2, y_pred_RF_tr2)
print("Random Forest Classification Model Score (Training Set): %.2f%%" % (accuracy_RF_tr2 * 100.0))
# testing set
y_pred_RF_t2 = Random_Forest_classifier.predict(X_test2)
predictions_RF2 = [round(value) for value in y_pred_RF_t2]
# evaluate predictions
accuracy_RF_t2 = accuracy_score(y_test2, predictions_RF2)
print("Random Forest Classification Model Score (Testing Set): %.2f%%" % (accuracy_RF_t2 * 100.0))
Random Forest Classification Model Score (Training Set): 88.36% Random Forest Classification Model Score (Testing Set): 86.12%
plot_confusion_matrix(Random_Forest_classifier, X_test2, y_test2,cmap="YlGnBu");
# Printing the classification report
print(classification_report(y_test2, y_pred_RF_t2))
precision recall f1-score support
0 0.98 0.87 0.92 468
1 0.26 0.72 0.38 29
accuracy 0.86 497
macro avg 0.62 0.80 0.65 497
weighted avg 0.94 0.86 0.89 497
Random_Forest_classifier.fit(X_train3, y_train3)
RandomForestClassifier(class_weight='balanced', min_samples_split=100,
n_estimators=300, random_state=42)In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. RandomForestClassifier(class_weight='balanced', min_samples_split=100,
n_estimators=300, random_state=42)# training set
y_pred_RF_tr3 = Random_Forest_classifier.predict(X_train3)
accuracy_RF_tr3 = accuracy_score(y_train3, y_pred_RF_tr3)
print("Random Forest Classification Model Score (Training Set): %.2f%%" % (accuracy_RF_tr3 * 100.0))
# testing set
y_pred_RF_t3 = Random_Forest_classifier.predict(X_test3)
predictions_RF3 = [round(value) for value in y_pred_RF_t3]
# evaluate predictions
accuracy_RF_t3 = accuracy_score(y_test3, predictions_RF3)
print("Random Forest Classification Model Score (Testing Set): %.2f%%" % (accuracy_RF_t3 * 100.0))
Random Forest Classification Model Score (Training Set): 97.43% Random Forest Classification Model Score (Testing Set): 93.76%
plot_confusion_matrix(Random_Forest_classifier, X_test3, y_test3,cmap="YlGnBu");
# Printing the classification report
print(classification_report(y_test3, y_pred_RF_t3))
precision recall f1-score support
0 0.00 0.00 0.00 17
1 0.96 0.97 0.97 480
accuracy 0.94 497
macro avg 0.48 0.49 0.48 497
weighted avg 0.93 0.94 0.93 497
Random_Forest_classifier.fit(X_train4, y_train4)
RandomForestClassifier(class_weight='balanced', min_samples_split=100,
n_estimators=300, random_state=42)In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. RandomForestClassifier(class_weight='balanced', min_samples_split=100,
n_estimators=300, random_state=42)# training set
y_pred_RF_tr4 = Random_Forest_classifier.predict(X_train4)
accuracy_RF_tr4 = accuracy_score(y_train4, y_pred_RF_tr4)
print("Random Forest Classification Model Score (Training Set): %.2f%%" % (accuracy_RF_tr4 * 100.0))
# testing set
y_pred_RF_t4 = Random_Forest_classifier.predict(X_test4)
predictions_RF4 = [round(value) for value in y_pred_RF_t4]
# evaluate predictions
accuracy_RF_t4 = accuracy_score(y_test4, predictions_RF4)
print("Random Forest Classification Model Score (Testing Set): %.2f%%" % (accuracy_RF_t4 * 100.0))
Random Forest Classification Model Score (Training Set): 74.40% Random Forest Classification Model Score (Testing Set): 67.61%
plot_confusion_matrix(Random_Forest_classifier, X_test4, y_test4,cmap="YlGnBu");
# Printing the classification report
print(classification_report(y_test4, y_pred_RF_t4))
precision recall f1-score support
0 0.64 0.62 0.63 222
1 0.70 0.72 0.71 275
accuracy 0.68 497
macro avg 0.67 0.67 0.67 497
weighted avg 0.68 0.68 0.68 497
Random_Forest_classifier.fit(X_train5, y_train5)
RandomForestClassifier(class_weight='balanced', min_samples_split=100,
n_estimators=300, random_state=42)In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. RandomForestClassifier(class_weight='balanced', min_samples_split=100,
n_estimators=300, random_state=42)# training set
y_pred_RF_tr5 = Random_Forest_classifier.predict(X_train5)
accuracy_RF_tr5 = accuracy_score(y_train5, y_pred_RF_tr5)
print("Random Forest Classification Model Score (Training Set): %.2f%%" % (accuracy_RF_tr5 * 100.0))
# testing set
y_pred_RF_t5 = Random_Forest_classifier.predict(X_test5)
predictions_RF5 = [round(value) for value in y_pred_RF_t5]
# evaluate predictions
accuracy_RF_t5 = accuracy_score(y_test5, predictions_RF5)
print("Random Forest Classification Model Score (Testing Set): %.2f%%" % (accuracy_RF_t5 * 100.0))
Random Forest Classification Model Score (Training Set): 76.27% Random Forest Classification Model Score (Testing Set): 73.84%
plot_confusion_matrix(Random_Forest_classifier, X_test5, y_test5,cmap="YlGnBu");
# Printing the classification report
print(classification_report(y_test5, y_pred_RF_t5))
precision recall f1-score support
0 0.84 0.73 0.78 318
1 0.61 0.76 0.68 179
accuracy 0.74 497
macro avg 0.73 0.74 0.73 497
weighted avg 0.76 0.74 0.74 497
# Accuracy Score for each company.
RF_accuracy = {'Aramco': accuracy_RF_t1*100, 'SARCO': accuracy_RF_t2*100, 'Rabigh': accuracy_RF_t3*100, 'Bahri':accuracy_RF_t4*100, 'Aldrees': accuracy_RF_t5*100}
print(RF_accuracy)
RF_names = list(RF_accuracy.keys())
RF_values = list(RF_accuracy.values())
fig, ax = plt.subplots(figsize=(15,8))
plt.title('Random Forest Classification Accuracy Score per Company')
RF_bars = ax.barh(RF_names, RF_values)
ax.bar_label(RF_bars, padding=5)
plt.xlim([0,100])
plt.show()
{'Aramco': 64.92537313432835, 'SARCO': 86.11670020120724, 'Rabigh': 93.7625754527163, 'Bahri': 67.6056338028169, 'Aldrees': 73.8430583501006}
# Number of trees in random forest
n_estimators = [200]
# Number of features to consider at every split
max_features = ['auto', 'sqrt', None, 'log2']
# Maximum number of levels in tree
max_depth = [40]
max_depth.append(None)
# Minimum number of samples required to split a node
min_samples_split = [1, 10, 20]
# Minimum number of samples required at each leaf node
min_samples_leaf = [1, 14, 16 ,20]
# Method of selecting samples for training each tree
bootstrap = [True, False]
# Create the random grid
random_grid = {'n_estimators': n_estimators,
'max_features': max_features,
'max_depth': max_depth,
'min_samples_split': min_samples_split,
'min_samples_leaf': min_samples_leaf,
'bootstrap': bootstrap}
print(random_grid)
{'n_estimators': [200], 'max_features': ['auto', 'sqrt', None, 'log2'], 'max_depth': [40, None], 'min_samples_split': [1, 10, 20], 'min_samples_leaf': [1, 14, 16, 20], 'bootstrap': [True, False]}
# split the dataset into Training and Testing sets
X_trainRF1, X_testRF1, y_trainRF1, y_testRF1 = split_data(df_Aramco.drop('SMA_20', axis=1), 'Target')
X_trainRF2, X_testRF2, y_trainRF2, y_testRF2 = split_data(df_SARCO.drop('SMA_20', axis=1), 'Target')
X_trainRF3, X_testRF3, y_trainRF3, y_testRF3 = split_data(df_Rabigh.drop('SMA_20', axis=1), 'Target')
X_trainRF4, X_testRF4, y_trainRF4, y_testRF4 = split_data(df_Bahri.drop('SMA_20', axis=1), 'Target')
X_trainRF5, X_testRF5, y_trainRF5, y_testRF5 = split_data(df_Aldrees.drop('SMA_20', axis=1), 'Target')
# New Random Forest Classifier to house optimal parameters
Random_Forest_classifier2 = RandomForestClassifier(class_weight="balanced")
# Specfiy the details of our Randomized Search
Random_Forest_classifier2_random = RandomizedSearchCV(estimator = Random_Forest_classifier2, param_distributions = random_grid, n_iter = 100, cv = 10, verbose=2, random_state=42, n_jobs = -1)
# Fit the random search model
Random_Forest_classifier2_random.fit(X_trainRF1, y_trainRF1)
Fitting 10 folds for each of 100 candidates, totalling 1000 fits
RandomizedSearchCV(cv=10,
estimator=RandomForestClassifier(class_weight='balanced'),
n_iter=100, n_jobs=-1,
param_distributions={'bootstrap': [True, False],
'max_depth': [40, None],
'max_features': ['auto', 'sqrt', None,
'log2'],
'min_samples_leaf': [1, 14, 16, 20],
'min_samples_split': [1, 10, 20],
'n_estimators': [200]},
random_state=42, verbose=2)In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. RandomizedSearchCV(cv=10,
estimator=RandomForestClassifier(class_weight='balanced'),
n_iter=100, n_jobs=-1,
param_distributions={'bootstrap': [True, False],
'max_depth': [40, None],
'max_features': ['auto', 'sqrt', None,
'log2'],
'min_samples_leaf': [1, 14, 16, 20],
'min_samples_split': [1, 10, 20],
'n_estimators': [200]},
random_state=42, verbose=2)RandomForestClassifier(class_weight='balanced')
RandomForestClassifier(class_weight='balanced')
print('\n Best estimator:')
print(Random_Forest_classifier2_random.best_estimator_)
print('\n Best score ')
print(Random_Forest_classifier2_random.best_score_ )
print('\n Best hyperparameters:')
print(Random_Forest_classifier2_random.best_params_)
Best estimator:
RandomForestClassifier(class_weight='balanced', max_features=None,
min_samples_leaf=20, min_samples_split=10,
n_estimators=200)
Best score
0.6342068483577917
Best hyperparameters:
{'n_estimators': 200, 'min_samples_split': 10, 'min_samples_leaf': 20, 'max_features': None, 'max_depth': None, 'bootstrap': True}
# training set
y_pred_RF_tr11 = Random_Forest_classifier2_random.predict(X_trainRF1)
accuracy_RF_tr11 = accuracy_score(y_trainRF1, y_pred_RF_tr11)
print("Random Forest Classification Model Score (Training Set): %.2f%%" % (accuracy_RF_tr11 * 100.0))
# testing set
y_pred_RF_t11 = Random_Forest_classifier2_random.predict(X_testRF1)
predictions_RF11 = [round(value) for value in y_pred_RF_t11]
# evaluate predictions
accuracy_RF_t11 = accuracy_score(y_testRF1, predictions_RF11)
print("Random Forest Classification Model Score (Testing Set): %.2f%%" % (accuracy_RF_t11 * 100.0))
Random Forest Classification Model Score (Training Set): 73.88% Random Forest Classification Model Score (Testing Set): 67.91%
plot_confusion_matrix(Random_Forest_classifier2_random, X_testRF1, y_testRF1,cmap="YlGnBu");
# Printing the classification report
print(classification_report(y_testRF1, y_pred_RF_t11))
precision recall f1-score support
0 0.59 0.69 0.63 54
1 0.76 0.68 0.72 80
accuracy 0.68 134
macro avg 0.67 0.68 0.67 134
weighted avg 0.69 0.68 0.68 134
Random_Forest_classifier2_random.fit(X_trainRF2, y_trainRF2)
Fitting 10 folds for each of 100 candidates, totalling 1000 fits
RandomizedSearchCV(cv=10,
estimator=RandomForestClassifier(class_weight='balanced'),
n_iter=100, n_jobs=-1,
param_distributions={'bootstrap': [True, False],
'max_depth': [40, None],
'max_features': ['auto', 'sqrt', None,
'log2'],
'min_samples_leaf': [1, 14, 16, 20],
'min_samples_split': [1, 10, 20],
'n_estimators': [200]},
random_state=42, verbose=2)In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. RandomizedSearchCV(cv=10,
estimator=RandomForestClassifier(class_weight='balanced'),
n_iter=100, n_jobs=-1,
param_distributions={'bootstrap': [True, False],
'max_depth': [40, None],
'max_features': ['auto', 'sqrt', None,
'log2'],
'min_samples_leaf': [1, 14, 16, 20],
'min_samples_split': [1, 10, 20],
'n_estimators': [200]},
random_state=42, verbose=2)RandomForestClassifier(class_weight='balanced')
RandomForestClassifier(class_weight='balanced')
# training set
y_pred_RF_tr22 = Random_Forest_classifier2_random.predict(X_trainRF2)
accuracy_RF_tr22 = accuracy_score(y_trainRF2, y_pred_RF_tr22)
print("Random Forest Classification Model Score (Training Set): %.2f%%" % (accuracy_RF_tr22 * 100.0))
# testing set
y_pred_RF_t22 = Random_Forest_classifier2_random.predict(X_testRF2)
predictions_RF22 = [round(value) for value in y_pred_RF_t22]
# evaluate predictions
accuracy_RF_t22 = accuracy_score(y_testRF2, predictions_RF22)
print("Random Forest Classification Model Score (Testing Set): %.2f%%" % (accuracy_RF_t22 * 100.0))
Random Forest Classification Model Score (Training Set): 99.90% Random Forest Classification Model Score (Testing Set): 93.76%
plot_confusion_matrix(Random_Forest_classifier2_random, X_testRF2, y_testRF2,cmap="YlGnBu");
# Printing the classification report
print(classification_report(y_testRF2, y_pred_RF_t22))
precision recall f1-score support
0 0.95 0.99 0.97 468
1 0.42 0.17 0.24 29
accuracy 0.94 497
macro avg 0.68 0.58 0.61 497
weighted avg 0.92 0.94 0.93 497
Random_Forest_classifier2_random.fit(X_trainRF3, y_trainRF3)
Fitting 10 folds for each of 100 candidates, totalling 1000 fits
RandomizedSearchCV(cv=10,
estimator=RandomForestClassifier(class_weight='balanced'),
n_iter=100, n_jobs=-1,
param_distributions={'bootstrap': [True, False],
'max_depth': [40, None],
'max_features': ['auto', 'sqrt', None,
'log2'],
'min_samples_leaf': [1, 14, 16, 20],
'min_samples_split': [1, 10, 20],
'n_estimators': [200]},
random_state=42, verbose=2)In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. RandomizedSearchCV(cv=10,
estimator=RandomForestClassifier(class_weight='balanced'),
n_iter=100, n_jobs=-1,
param_distributions={'bootstrap': [True, False],
'max_depth': [40, None],
'max_features': ['auto', 'sqrt', None,
'log2'],
'min_samples_leaf': [1, 14, 16, 20],
'min_samples_split': [1, 10, 20],
'n_estimators': [200]},
random_state=42, verbose=2)RandomForestClassifier(class_weight='balanced')
RandomForestClassifier(class_weight='balanced')
# training set
y_pred_RF_tr33 = Random_Forest_classifier2_random.predict(X_trainRF3)
accuracy_RF_tr33 = accuracy_score(y_trainRF3, y_pred_RF_tr33)
print("Random Forest Classification Model Score (Training Set): %.2f%%" % (accuracy_RF_tr33 * 100.0))
# testing set
y_pred_RF_t33 = Random_Forest_classifier2_random.predict(X_testRF3)
predictions_RF33 = [round(value) for value in y_pred_RF_t33]
# evaluate predictions
accuracy_RF_t33 = accuracy_score(y_testRF3, predictions_RF33)
print("Random Forest Classification Model Score (Testing Set): %.2f%%" % (accuracy_RF_t33 * 100.0))
Random Forest Classification Model Score (Training Set): 100.00% Random Forest Classification Model Score (Testing Set): 96.58%
plot_confusion_matrix(Random_Forest_classifier2_random, X_testRF3, y_testRF3,cmap="YlGnBu");
# Printing the classification report
print(classification_report(y_testRF3, y_pred_RF_t33))
precision recall f1-score support
0 0.00 0.00 0.00 17
1 0.97 1.00 0.98 480
accuracy 0.97 497
macro avg 0.48 0.50 0.49 497
weighted avg 0.93 0.97 0.95 497
Random_Forest_classifier2_random.fit(X_trainRF4, y_trainRF4)
Fitting 10 folds for each of 100 candidates, totalling 1000 fits
RandomizedSearchCV(cv=10,
estimator=RandomForestClassifier(class_weight='balanced'),
n_iter=100, n_jobs=-1,
param_distributions={'bootstrap': [True, False],
'max_depth': [40, None],
'max_features': ['auto', 'sqrt', None,
'log2'],
'min_samples_leaf': [1, 14, 16, 20],
'min_samples_split': [1, 10, 20],
'n_estimators': [200]},
random_state=42, verbose=2)In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. RandomizedSearchCV(cv=10,
estimator=RandomForestClassifier(class_weight='balanced'),
n_iter=100, n_jobs=-1,
param_distributions={'bootstrap': [True, False],
'max_depth': [40, None],
'max_features': ['auto', 'sqrt', None,
'log2'],
'min_samples_leaf': [1, 14, 16, 20],
'min_samples_split': [1, 10, 20],
'n_estimators': [200]},
random_state=42, verbose=2)RandomForestClassifier(class_weight='balanced')
RandomForestClassifier(class_weight='balanced')
# training set
y_pred_RF_tr44 = Random_Forest_classifier2_random.predict(X_trainRF4)
accuracy_RF_tr44 = accuracy_score(y_trainRF4, y_pred_RF_tr44)
print("Random Forest Classification Model Score (Training Set): %.2f%%" % (accuracy_RF_tr44 * 100.0))
# testing set
y_pred_RF_t44 = Random_Forest_classifier2_random.predict(X_testRF4)
predictions_RF44 = [round(value) for value in y_pred_RF_t44]
# evaluate predictions
accuracy_RF_t44 = accuracy_score(y_testRF4, predictions_RF44)
print("Random Forest Classification Model Score (Testing Set): %.2f%%" % (accuracy_RF_t44 * 100.0))
Random Forest Classification Model Score (Training Set): 89.92% Random Forest Classification Model Score (Testing Set): 68.61%
plot_confusion_matrix(Random_Forest_classifier2_random, X_testRF4, y_testRF4,cmap="YlGnBu");
# Printing the classification report
print(classification_report(y_testRF4, y_pred_RF_t44))
precision recall f1-score support
0 0.66 0.61 0.64 222
1 0.70 0.75 0.72 275
accuracy 0.69 497
macro avg 0.68 0.68 0.68 497
weighted avg 0.68 0.69 0.68 497
Random_Forest_classifier2_random.fit(X_trainRF5, y_trainRF5)
Fitting 10 folds for each of 100 candidates, totalling 1000 fits
RandomizedSearchCV(cv=10,
estimator=RandomForestClassifier(class_weight='balanced'),
n_iter=100, n_jobs=-1,
param_distributions={'bootstrap': [True, False],
'max_depth': [40, None],
'max_features': ['auto', 'sqrt', None,
'log2'],
'min_samples_leaf': [1, 14, 16, 20],
'min_samples_split': [1, 10, 20],
'n_estimators': [200]},
random_state=42, verbose=2)In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. RandomizedSearchCV(cv=10,
estimator=RandomForestClassifier(class_weight='balanced'),
n_iter=100, n_jobs=-1,
param_distributions={'bootstrap': [True, False],
'max_depth': [40, None],
'max_features': ['auto', 'sqrt', None,
'log2'],
'min_samples_leaf': [1, 14, 16, 20],
'min_samples_split': [1, 10, 20],
'n_estimators': [200]},
random_state=42, verbose=2)RandomForestClassifier(class_weight='balanced')
RandomForestClassifier(class_weight='balanced')
# training set
y_pred_RF_tr55 = Random_Forest_classifier2_random.predict(X_trainRF5)
accuracy_RF_tr55 = accuracy_score(y_trainRF5, y_pred_RF_tr55)
print("Random Forest Classification Model Score (Training Set): %.2f%%" % (accuracy_RF_tr55 * 100.0))
# testing set
y_pred_RF_t55 = Random_Forest_classifier2_random.predict(X_testRF5)
predictions_RF55 = [round(value) for value in y_pred_RF_t55]
# evaluate predictions
accuracy_RF_t55 = accuracy_score(y_testRF5, predictions_RF55)
print("Random Forest Classification Model Score (Testing Set): %.2f%%" % (accuracy_RF_t55 * 100.0))
Random Forest Classification Model Score (Training Set): 97.18% Random Forest Classification Model Score (Testing Set): 75.05%
plot_confusion_matrix(Random_Forest_classifier2_random, X_testRF5, y_testRF5,cmap="YlGnBu");
# Printing the classification report
print(classification_report(y_testRF5, y_pred_RF_t55))
precision recall f1-score support
0 0.79 0.82 0.81 318
1 0.66 0.62 0.64 179
accuracy 0.75 497
macro avg 0.73 0.72 0.73 497
weighted avg 0.75 0.75 0.75 497
# Accuracy Score for each company.
RF_accuracy2 = {'Aramco': accuracy_RF_t11*100, 'SARCO': accuracy_RF_t22*100, 'Rabigh': accuracy_RF_t33*100, 'Bahri':accuracy_RF_t44*100, 'Aldrees': accuracy_RF_t55*100}
RF_names2 = list(RF_accuracy2.keys())
RF_values2 = list(RF_accuracy2.values())
print(RF_accuracy2)
fig, ax = plt.subplots(figsize=(15,8))
plt.title('Random Forest Classification using RandomSearch CV Classification Accuracy Score per Company')
RF_bars2 = ax.barh(RF_names2, RF_values2)
ax.bar_label(RF_bars2, padding=5)
plt.xlim([0,100])
plt.show()
{'Aramco': 67.91044776119402, 'SARCO': 93.7625754527163, 'Rabigh': 96.579476861167, 'Bahri': 68.61167002012073, 'Aldrees': 75.0503018108652}
param_grid = {
'bootstrap': [True],
'max_depth': [80],
'max_features': [3],
'min_samples_leaf': [5],
'min_samples_split': [8],
'n_estimators': [200]
}
RF_grid = GridSearchCV(
Random_Forest_classifier,
param_grid,
cv = 10,
n_jobs=-1,
verbose=2
)
# Fit the random search model
RF_grid.fit(X_trainRF1, y_trainRF1)
Fitting 10 folds for each of 1 candidates, totalling 10 fits
GridSearchCV(cv=10,
estimator=RandomForestClassifier(class_weight='balanced',
min_samples_split=100,
n_estimators=300,
random_state=42),
n_jobs=-1,
param_grid={'bootstrap': [True], 'max_depth': [80],
'max_features': [3], 'min_samples_leaf': [5],
'min_samples_split': [8], 'n_estimators': [200]},
verbose=2)In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. GridSearchCV(cv=10,
estimator=RandomForestClassifier(class_weight='balanced',
min_samples_split=100,
n_estimators=300,
random_state=42),
n_jobs=-1,
param_grid={'bootstrap': [True], 'max_depth': [80],
'max_features': [3], 'min_samples_leaf': [5],
'min_samples_split': [8], 'n_estimators': [200]},
verbose=2)RandomForestClassifier(class_weight='balanced', min_samples_split=100,
n_estimators=300, random_state=42)RandomForestClassifier(class_weight='balanced', min_samples_split=100,
n_estimators=300, random_state=42)# training set
y_pred_RF_tr111 = RF_grid.predict(X_trainRF1)
accuracy_RF_tr111 = accuracy_score(y_trainRF1, y_pred_RF_tr111)
print("Random Forest Classification Model Score (Training Set): %.2f%%" % (accuracy_RF_tr111 * 100.0))
# testing set
y_pred_RF_t111 = RF_grid.predict(X_testRF1)
predictions_RF111 = [round(value) for value in y_pred_RF_t111]
# evaluate predictions
accuracy_RF_t111 = accuracy_score(y_testRF1, predictions_RF111)
print("Random Forest Classification Model Score (Testing Set): %.2f%%" % (accuracy_RF_t111 * 100.0))
Random Forest Classification Model Score (Training Set): 91.60% Random Forest Classification Model Score (Testing Set): 68.66%
plot_confusion_matrix(RF_grid, X_testRF1, y_testRF1,cmap="YlGnBu");
# Printing the classification report
print(classification_report(y_testRF1, y_pred_RF_t111))
precision recall f1-score support
0 0.62 0.59 0.60 54
1 0.73 0.75 0.74 80
accuracy 0.69 134
macro avg 0.67 0.67 0.67 134
weighted avg 0.68 0.69 0.69 134
RF_grid.fit(X_trainRF2, y_trainRF2)
Fitting 10 folds for each of 1 candidates, totalling 10 fits
GridSearchCV(cv=10,
estimator=RandomForestClassifier(class_weight='balanced',
min_samples_split=100,
n_estimators=300,
random_state=42),
n_jobs=-1,
param_grid={'bootstrap': [True], 'max_depth': [80],
'max_features': [3], 'min_samples_leaf': [5],
'min_samples_split': [8], 'n_estimators': [200]},
verbose=2)In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. GridSearchCV(cv=10,
estimator=RandomForestClassifier(class_weight='balanced',
min_samples_split=100,
n_estimators=300,
random_state=42),
n_jobs=-1,
param_grid={'bootstrap': [True], 'max_depth': [80],
'max_features': [3], 'min_samples_leaf': [5],
'min_samples_split': [8], 'n_estimators': [200]},
verbose=2)RandomForestClassifier(class_weight='balanced', min_samples_split=100,
n_estimators=300, random_state=42)RandomForestClassifier(class_weight='balanced', min_samples_split=100,
n_estimators=300, random_state=42)# training set
y_pred_RF_tr222 = RF_grid.predict(X_trainRF2)
accuracy_RF_tr222 = accuracy_score(y_trainRF2, y_pred_RF_tr222)
print("Random Forest Classification Model Score (Training Set): %.2f%%" % (accuracy_RF_tr222 * 100.0))
# testing set
y_pred_RF_t222 = RF_grid.predict(X_testRF2)
predictions_RF222 = [round(value) for value in y_pred_RF_t222]
# evaluate predictions
accuracy_RF_t222 = accuracy_score(y_testRF2, predictions_RF222)
print("Random Forest Classification Model Score (Testing Set): %.2f%%" % (accuracy_RF_t222 * 100.0))
Random Forest Classification Model Score (Training Set): 98.34% Random Forest Classification Model Score (Testing Set): 93.36%
plot_confusion_matrix(RF_grid, X_testRF2, y_testRF2,cmap="YlGnBu");
# Printing the classification report
print(classification_report(y_testRF2, y_pred_RF_t222))
precision recall f1-score support
0 0.97 0.96 0.96 468
1 0.43 0.45 0.44 29
accuracy 0.93 497
macro avg 0.70 0.71 0.70 497
weighted avg 0.93 0.93 0.93 497
RF_grid.fit(X_trainRF3, y_trainRF3)
Fitting 10 folds for each of 1 candidates, totalling 10 fits
GridSearchCV(cv=10,
estimator=RandomForestClassifier(class_weight='balanced',
min_samples_split=100,
n_estimators=300,
random_state=42),
n_jobs=-1,
param_grid={'bootstrap': [True], 'max_depth': [80],
'max_features': [3], 'min_samples_leaf': [5],
'min_samples_split': [8], 'n_estimators': [200]},
verbose=2)In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. GridSearchCV(cv=10,
estimator=RandomForestClassifier(class_weight='balanced',
min_samples_split=100,
n_estimators=300,
random_state=42),
n_jobs=-1,
param_grid={'bootstrap': [True], 'max_depth': [80],
'max_features': [3], 'min_samples_leaf': [5],
'min_samples_split': [8], 'n_estimators': [200]},
verbose=2)RandomForestClassifier(class_weight='balanced', min_samples_split=100,
n_estimators=300, random_state=42)RandomForestClassifier(class_weight='balanced', min_samples_split=100,
n_estimators=300, random_state=42)# training set
y_pred_RF_tr333 = RF_grid.predict(X_trainRF3)
accuracy_RF_tr333 = accuracy_score(y_trainRF3, y_pred_RF_tr333)
print("Random Forest Classification Model Score (Training Set): %.2f%%" % (accuracy_RF_tr333 * 100.0))
# testing set
y_pred_RF_t333 = RF_grid.predict(X_testRF3)
predictions_RF333 = [round(value) for value in y_pred_RF_t333]
# evaluate predictions
accuracy_RF_t333 = accuracy_score(y_testRF3, predictions_RF333)
print("Random Forest Classification Model Score (Testing Set): %.2f%%" % (accuracy_RF_t333 * 100.0))
Random Forest Classification Model Score (Training Set): 99.90% Random Forest Classification Model Score (Testing Set): 96.58%
plot_confusion_matrix(RF_grid, X_testRF3, y_testRF3,cmap="YlGnBu");
# Printing the classification report
print(classification_report(y_testRF3, y_pred_RF_t333))
precision recall f1-score support
0 0.00 0.00 0.00 17
1 0.97 1.00 0.98 480
accuracy 0.97 497
macro avg 0.48 0.50 0.49 497
weighted avg 0.93 0.97 0.95 497
RF_grid.fit(X_trainRF4, y_trainRF4)
Fitting 10 folds for each of 1 candidates, totalling 10 fits
GridSearchCV(cv=10,
estimator=RandomForestClassifier(class_weight='balanced',
min_samples_split=100,
n_estimators=300,
random_state=42),
n_jobs=-1,
param_grid={'bootstrap': [True], 'max_depth': [80],
'max_features': [3], 'min_samples_leaf': [5],
'min_samples_split': [8], 'n_estimators': [200]},
verbose=2)In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. GridSearchCV(cv=10,
estimator=RandomForestClassifier(class_weight='balanced',
min_samples_split=100,
n_estimators=300,
random_state=42),
n_jobs=-1,
param_grid={'bootstrap': [True], 'max_depth': [80],
'max_features': [3], 'min_samples_leaf': [5],
'min_samples_split': [8], 'n_estimators': [200]},
verbose=2)RandomForestClassifier(class_weight='balanced', min_samples_split=100,
n_estimators=300, random_state=42)RandomForestClassifier(class_weight='balanced', min_samples_split=100,
n_estimators=300, random_state=42)# training set
y_pred_RF_tr444 = RF_grid.predict(X_trainRF4)
accuracy_RF_tr444 = accuracy_score(y_trainRF4, y_pred_RF_tr444)
print("Random Forest Classification Model Score (Training Set): %.2f%%" % (accuracy_RF_tr44 * 100.0))
# testing set
y_pred_RF_t444 = RF_grid.predict(X_testRF4)
predictions_RF444 = [round(value) for value in y_pred_RF_t444]
# evaluate predictions
accuracy_RF_t444 = accuracy_score(y_testRF4, predictions_RF444)
print("Random Forest Classification Model Score (Testing Set): %.2f%%" % (accuracy_RF_t444 * 100.0))
Random Forest Classification Model Score (Training Set): 89.92% Random Forest Classification Model Score (Testing Set): 67.00%
plot_confusion_matrix(RF_grid, X_testRF4, y_testRF4,cmap="YlGnBu");
# Printing the classification report
print(classification_report(y_testRF4, y_pred_RF_t444))
precision recall f1-score support
0 0.65 0.57 0.61 222
1 0.68 0.75 0.72 275
accuracy 0.67 497
macro avg 0.67 0.66 0.66 497
weighted avg 0.67 0.67 0.67 497
RF_grid.fit(X_trainRF5, y_trainRF5)
Fitting 10 folds for each of 1 candidates, totalling 10 fits
GridSearchCV(cv=10,
estimator=RandomForestClassifier(class_weight='balanced',
min_samples_split=100,
n_estimators=300,
random_state=42),
n_jobs=-1,
param_grid={'bootstrap': [True], 'max_depth': [80],
'max_features': [3], 'min_samples_leaf': [5],
'min_samples_split': [8], 'n_estimators': [200]},
verbose=2)In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. GridSearchCV(cv=10,
estimator=RandomForestClassifier(class_weight='balanced',
min_samples_split=100,
n_estimators=300,
random_state=42),
n_jobs=-1,
param_grid={'bootstrap': [True], 'max_depth': [80],
'max_features': [3], 'min_samples_leaf': [5],
'min_samples_split': [8], 'n_estimators': [200]},
verbose=2)RandomForestClassifier(class_weight='balanced', min_samples_split=100,
n_estimators=300, random_state=42)RandomForestClassifier(class_weight='balanced', min_samples_split=100,
n_estimators=300, random_state=42)# training set
y_pred_RF_tr555 = RF_grid.predict(X_trainRF5)
accuracy_RF_tr555 = accuracy_score(y_trainRF5, y_pred_RF_tr555)
print("Random Forest Classification Model Score (Training Set): %.2f%%" % (accuracy_RF_tr555 * 100.0))
# testing set
y_pred_RF_t555 = RF_grid.predict(X_testRF5)
predictions_RF555 = [round(value) for value in y_pred_RF_t555]
# evaluate predictions
accuracy_RF_t555 = accuracy_score(y_testRF5, predictions_RF555)
print("Random Forest Classification Model Score (Testing Set): %.2f%%" % (accuracy_RF_t555 * 100.0))
Random Forest Classification Model Score (Training Set): 92.49% Random Forest Classification Model Score (Testing Set): 74.45%
plot_confusion_matrix(RF_grid, X_testRF5, y_testRF5,cmap="YlGnBu");
# Printing the classification report
print(classification_report(y_testRF5, y_pred_RF_t555))
precision recall f1-score support
0 0.81 0.79 0.80 318
1 0.64 0.66 0.65 179
accuracy 0.74 497
macro avg 0.72 0.73 0.73 497
weighted avg 0.75 0.74 0.75 497
# Accuracy Score for each company.
RF_accuracy3 = {'Aramco': accuracy_RF_t111*100, 'SARCO': accuracy_RF_t222*100, 'Rabigh': accuracy_RF_t333*100,\
'Bahri':accuracy_RF_t444*100, 'Aldrees': accuracy_RF_t555*100}
RF_names3 = list(RF_accuracy3.keys())
RF_values3 = list(RF_accuracy3.values())
print(RF_accuracy3)
fig, ax = plt.subplots(figsize=(15,8))
plt.title('Random Forest Classification using GridSearch CV Classification Accuracy Score per Company')
RF_bars3 = ax.barh(RF_names3, RF_values3)
ax.bar_label(RF_bars3, padding=5)
plt.xlim([0,100])
plt.show()
{'Aramco': 68.65671641791045, 'SARCO': 93.36016096579478, 'Rabigh': 96.579476861167, 'Bahri': 67.0020120724346, 'Aldrees': 74.44668008048289}
df_RF = pd.DataFrame({
'Company': RF_names,
'Random_Forest': RF_values,
'Random_Forest_GridSearch': RF_values2,
'Random_Forest_RanomizedSearch': RF_values3
})
df_RF
| Company | Random_Forest | Random_Forest_GridSearch | Random_Forest_RanomizedSearch | |
|---|---|---|---|---|
| 0 | Aramco | 64.925373 | 67.910448 | 68.656716 |
| 1 | SARCO | 86.116700 | 93.762575 | 93.360161 |
| 2 | Rabigh | 93.762575 | 96.579477 | 96.579477 |
| 3 | Bahri | 67.605634 | 68.611670 | 67.002012 |
| 4 | Aldrees | 73.843058 | 75.050302 | 74.446680 |
df_RF.to_csv('RF_Accuracy.csv')
# Calculate log loss (error) for all the Random_Forest Models
RF_log_loss = {
'Aramco': log_loss(y_true=y_test1, y_pred=y_pred_RF_t1, eps=1e-15, normalize=True, sample_weight=None, labels=None),
'SARCO': log_loss(y_true=y_test2, y_pred=y_pred_RF_t2, eps=1e-15, normalize=True, sample_weight=None, labels=None),
'Rabigh': log_loss(y_true=y_test3, y_pred=y_pred_RF_t3, eps=1e-15, normalize=True, sample_weight=None, labels=None),
'Bahri': log_loss(y_true=y_test4, y_pred=y_pred_RF_t4, eps=1e-15, normalize=True, sample_weight=None, labels=None),
'Aldrees': log_loss(y_true=y_test5, y_pred=y_pred_RF_t5, eps=1e-15, normalize=True, sample_weight=None, labels=None)
}
RF_log_loss_values = list(RF_log_loss.values())
RF_gridsearch_log_loss = {
'Aramco': log_loss(y_true=y_test1, y_pred=y_pred_RF_t111, eps=1e-15, normalize=True, sample_weight=None, labels=None),
'SARCO': log_loss(y_true=y_test2, y_pred=y_pred_RF_t222, eps=1e-15, normalize=True, sample_weight=None, labels=None),
'Rabigh': log_loss(y_true=y_test3, y_pred=y_pred_RF_t333, eps=1e-15, normalize=True, sample_weight=None, labels=None),
'Bahri': log_loss(y_true=y_test4, y_pred=y_pred_RF_t444, eps=1e-15, normalize=True, sample_weight=None, labels=None),
'Aldrees': log_loss(y_true=y_test5, y_pred=y_pred_RF_t555, eps=1e-15, normalize=True, sample_weight=None, labels=None)
}
RF_gridsearch_log_loss_values = list(RF_gridsearch_log_loss.values())
RF_randomsearch_log_loss = {
'Aramco': log_loss(y_true=y_test1, y_pred=y_pred_RF_t11, eps=1e-15, normalize=True, sample_weight=None, labels=None),
'SARCO': log_loss(y_true=y_test2, y_pred=y_pred_RF_t22, eps=1e-15, normalize=True, sample_weight=None, labels=None),
'Rabigh': log_loss(y_true=y_test3, y_pred=y_pred_RF_t33, eps=1e-15, normalize=True, sample_weight=None, labels=None),
'Bahri': log_loss(y_true=y_test4, y_pred=y_pred_RF_t44, eps=1e-15, normalize=True, sample_weight=None, labels=None),
'Aldrees': log_loss(y_true=y_test5, y_pred=y_pred_RF_t55, eps=1e-15, normalize=True, sample_weight=None, labels=None)
}
RF_randomsearch_log_loss_values = list(RF_randomsearch_log_loss.values())
df_RF_log_loss = pd.DataFrame({
'Company': RF_names,
'Random_Forest': RF_log_loss_values,
'Random_Forest_GridSearch': RF_gridsearch_log_loss_values,
'Random_Forest_RanomizedSearch': RF_randomsearch_log_loss_values
})
df_RF_log_loss
| Company | Random_Forest | Random_Forest_GridSearch | Random_Forest_RanomizedSearch | |
|---|---|---|---|---|
| 0 | Aramco | 12.114484 | 10.825718 | 11.083440 |
| 1 | SARCO | 4.795220 | 2.293347 | 2.154341 |
| 2 | Rabigh | 2.154357 | 1.181434 | 1.181434 |
| 3 | Bahri | 11.188753 | 11.397254 | 10.841283 |
| 4 | Aldrees | 9.034428 | 8.825912 | 8.617411 |
df_RF_log_loss.to_csv('RF_errors.csv')
RF_f1 = {
'Aramco': f1_score(y_true=y_test1, y_pred=y_pred_RF_t1, average='macro'),
'SARCO': f1_score(y_true=y_test2, y_pred=y_pred_RF_t2, average='macro'),
'Rabigh': f1_score(y_true=y_test3, y_pred=y_pred_RF_t3, average='macro'),
'Bahri': f1_score(y_true=y_test4, y_pred=y_pred_RF_t4, average='macro'),
'Aldrees': f1_score(y_true=y_test5, y_pred=y_pred_RF_t5, average='macro')
}
RF_f1_values = list(RF_f1.values())
RF_gridsearch_f1 = {
'Aramco': f1_score(y_true=y_testRF1, y_pred=y_pred_RF_t111, average='macro'),
'SARCO': f1_score(y_true=y_testRF2, y_pred=y_pred_RF_t222, average='macro'),
'Rabigh': f1_score(y_true=y_testRF3, y_pred=y_pred_RF_t333, average='macro'),
'Bahri': f1_score(y_true=y_testRF4, y_pred=y_pred_RF_t444, average='macro'),
'Aldrees': f1_score(y_true=y_testRF5, y_pred=y_pred_RF_t555, average='macro')
}
RF_gridsearch_f1_values = list(RF_gridsearch_f1.values())
RF_randomsearch_f1 = {
'Aramco': f1_score(y_true=y_testRF1, y_pred=y_pred_RF_t11, average='macro'),
'SARCO': f1_score(y_true=y_testRF2, y_pred=y_pred_RF_t22, average='macro'),
'Rabigh': f1_score(y_true=y_testRF3, y_pred=y_pred_RF_t33, average='macro'),
'Bahri': f1_score(y_true=y_testRF4, y_pred=y_pred_RF_t44, average='macro'),
'Aldrees': f1_score(y_true=y_testRF5, y_pred=y_pred_RF_t55, average='macro')
}
RF_randomsearch_f1_values = list(RF_randomsearch_f1.values())
df_RF_f1 = pd.DataFrame({
'Company': RF_names,
'Random_Forest': RF_f1_values,
'Random_Forest_GridSearch': RF_gridsearch_f1_values,
'Random_Forest_RanomizedSearch': RF_randomsearch_f1_values
})
df_RF_f1
| Company | Random_Forest | Random_Forest_GridSearch | Random_Forest_RanomizedSearch | |
|---|---|---|---|---|
| 0 | Aramco | 0.636605 | 0.672257 | 0.673855 |
| 1 | SARCO | 0.650118 | 0.702692 | 0.605687 |
| 2 | Rabigh | 0.483904 | 0.491300 | 0.491300 |
| 3 | Bahri | 0.671265 | 0.661467 | 0.679948 |
| 4 | Aldrees | 0.728511 | 0.725074 | 0.725130 |
df_RF_f1.to_csv('RF_f1_score.csv')
# split data
X_train1, X_test1, y_train1, y_test1 = split_data(df_Aramco, 'Target')
orig_feature_names = list(X_train1.columns)
#scaling the data
X_train1, X_test1 = scale_data(X_train1, X_test1)
X_train1 = pd.DataFrame(X_train1, columns=orig_feature_names)
X_test1 = pd.DataFrame(X_test1, columns=orig_feature_names)
# count examples in each class
counter = Counter(df_Aramco['Target'])
# estimate scale_pos_weight value
estimate1 = counter[0] / counter[1]
print('Estimate: %.3f' % estimate1)
Estimate: 0.763
xgb_classifier1 = XGBClassifier(random_state=42, scale_pos_weight=estimate1)
xgb_classifier1.fit(X_train1, y_train1)
xgb.plot_importance(xgb_classifier1)
plt.show()
X_train1.shape
(536, 13)
# training set
y_pred_xgb_tr1 = xgb_classifier1.predict(X_train1)
accuracy_xgb_tr1 = accuracy_score(y_train1, y_pred_xgb_tr1)
print("XGBoost Model Score (Training Set): %.2f%%" % (accuracy_xgb_tr1 * 100.0))
# testing set
y_pred_xgb_t1 = xgb_classifier1.predict(X_test1)
xgb_predictions1 = [round(value) for value in y_pred_xgb_t1]
# evaluate predictions
accuracy_xgb_t1 = accuracy_score(y_test1, xgb_predictions1)
print("XGBoost Model Score (Testing Set): %.2f%%" % (accuracy_xgb_t1 * 100.0))
XGBoost Model Score (Training Set): 100.00% XGBoost Model Score (Testing Set): 66.42%
plot_confusion_matrix(xgb_classifier1, X_test1, y_test1,cmap="YlGnBu"); # good improvement
# Printing the classification report
print(classification_report(y_test1, y_pred_xgb_t1))
precision recall f1-score support
0 0.58 0.57 0.58 54
1 0.72 0.72 0.72 80
accuracy 0.66 134
macro avg 0.65 0.65 0.65 134
weighted avg 0.66 0.66 0.66 134
counter = Counter(df_SARCO['Target'])
# estimate scale_pos_weight value
estimate2 = counter[0] / counter[1]
print('Estimate: %.3f' % estimate2)
Estimate: 16.110
# split data
X_train2, X_test2, y_train2, y_test2 = split_data(df_SARCO, 'Target')
orig_feature_names = list(X_train1.columns)
#scaling the data
X_train2, X_test2 = scale_data(X_train2, X_test2)
X_train2 = pd.DataFrame(X_train2, columns=orig_feature_names)
X_test2 = pd.DataFrame(X_test2, columns=orig_feature_names)
xgb_classifier2 = XGBClassifier(random_state=42, scale_pos_weight=estimate2)
xgb_classifier2.fit(X_train2, y_train2)
xgb.plot_importance(xgb_classifier2)
plt.show()
# training set
y_pred_xgb_tr2 = xgb_classifier2.predict(X_train2)
accuracy_xgb_tr2 = accuracy_score(y_train2, y_pred_xgb_tr2)
print("XGBoost Model Score (Training Set): %.2f%%" % (accuracy_xgb_tr2 * 100.0))
# testing set
y_pred_xgb_t2 = xgb_classifier2.predict(X_test2)
xgb_predictions2 = [round(value) for value in y_pred_xgb_t2]
# evaluate predictions
accuracy_xgb_t2 = accuracy_score(y_test2, xgb_predictions2)
print("XGBoost Model Score (Testing Set): %.2f%%" % (accuracy_xgb_t2 * 100.0))
XGBoost Model Score (Training Set): 100.00% XGBoost Model Score (Testing Set): 93.56%
plot_confusion_matrix(xgb_classifier2, X_test2, y_test2,cmap="YlGnBu"); # good improvement
# Printing the classification report
print(classification_report(y_test2, y_pred_xgb_t2))
precision recall f1-score support
0 0.96 0.98 0.97 468
1 0.42 0.28 0.33 29
accuracy 0.94 497
macro avg 0.69 0.63 0.65 497
weighted avg 0.92 0.94 0.93 497
counter = Counter(df_Rabigh['Target'])
# estimate scale_pos_weight value
estimate3 = counter[0] / counter[1]
print('Estimate: %.3f' % estimate3)
Estimate: 0.032
# split data
X_train3, X_test3, y_train3, y_test3 = split_data(df_Rabigh, 'Target')
orig_feature_names = list(X_train1.columns)
#scaling the data
X_train3, X_test3 = scale_data(X_train3, X_test3)
X_train3 = pd.DataFrame(X_train3, columns=orig_feature_names)
X_test3 = pd.DataFrame(X_test3, columns=orig_feature_names)
xgb_classifier3 = XGBClassifier(random_state=42, scale_pos_weight=estimate3)
xgb_classifier3.fit(X_train3, y_train3)
xgb.plot_importance(xgb_classifier3)
plt.show()
# training set
y_pred_xgb_tr3 = xgb_classifier3.predict(X_train3)
accuracy_xgb_tr3 = accuracy_score(y_train3, y_pred_xgb_tr3)
print("XGBoost Model Score (Training Set): %.2f%%" % (accuracy_xgb_tr3 * 100.0))
# testing set
y_pred_xgb_t3 = xgb_classifier3.predict(X_test3)
xgb_predictions3 = [round(value) for value in y_pred_xgb_t3]
# evaluate predictions
accuracy_xgb_t3 = accuracy_score(y_test3, xgb_predictions3)
print("XGBoost Model Score (Testing Set): %.2f%%" % (accuracy_xgb_t3 * 100.0))
XGBoost Model Score (Training Set): 98.59% XGBoost Model Score (Testing Set): 93.96%
plot_confusion_matrix(xgb_classifier3, X_test3, y_test3,cmap="YlGnBu"); # good improvement
# Printing the classification report
print(classification_report(y_test3, y_pred_xgb_t3))
precision recall f1-score support
0 0.07 0.06 0.06 17
1 0.97 0.97 0.97 480
accuracy 0.94 497
macro avg 0.52 0.51 0.52 497
weighted avg 0.94 0.94 0.94 497
counter = Counter(df_Bahri['Target'])
# estimate scale_pos_weight value
estimate4 = counter[0] / counter[1]
print('Estimate: %.3f' % estimate4)
Estimate: 0.751
X_train4, X_test4, y_train4, y_test4 = split_data(df_Bahri, 'Target')
orig_feature_names = list(X_train1.columns)
#scaling the data
X_train4, X_test4 = scale_data(X_train4, X_test4)
X_train4 = pd.DataFrame(X_train4, columns=orig_feature_names)
X_test4 = pd.DataFrame(X_test4, columns=orig_feature_names)
xgb_classifier4 = XGBClassifier(random_state=42, scale_pos_weight=estimate4)
xgb_classifier4.fit(X_train4, y_train4)
xgb.plot_importance(xgb_classifier4)
plt.show()
# training set
y_pred_xgb_tr4 = xgb_classifier4.predict(X_train4)
accuracy_xgb_tr4 = accuracy_score(y_train4, y_pred_xgb_tr4)
print("XGBoost Model Score (Training Set): %.2f%%" % (accuracy_xgb_tr4 * 100.0))
# testing set
y_pred_xgb_t4 = xgb_classifier4.predict(X_test4)
xgb_predictions4 = [round(value) for value in y_pred_xgb_t4]
# evaluate predictions
accuracy_xgb_t4 = accuracy_score(y_test4, xgb_predictions4)
print("XGBoost Model Score (Testing Set): %.2f%%" % (accuracy_xgb_t4 * 100.0))
XGBoost Model Score (Training Set): 99.95% XGBoost Model Score (Testing Set): 64.99%
plot_confusion_matrix(xgb_classifier4, X_test4, y_test4,cmap="YlGnBu"); # good improvement
# Printing the classification report
print(classification_report(y_test4, y_pred_xgb_t4))
precision recall f1-score support
0 0.61 0.58 0.60 222
1 0.68 0.71 0.69 275
accuracy 0.65 497
macro avg 0.65 0.64 0.64 497
weighted avg 0.65 0.65 0.65 497
counter = Counter(df_Aldrees['Target'])
# estimate scale_pos_weight value
estimate5 = counter[0] / counter[1]
print('Estimate: %.3f' % estimate5)
Estimate: 1.749
X_train5, X_test5, y_train5, y_test5 = split_data(df_Aldrees, 'Target')
orig_feature_names = list(X_train1.columns)
#scaling the data
X_train5, X_test5 = scale_data(X_train5, X_test5)
X_train5 = pd.DataFrame(X_train5, columns=orig_feature_names)
X_test5 = pd.DataFrame(X_test5, columns=orig_feature_names)
xgb_classifier5 = XGBClassifier(random_state=42, scale_pos_weight=estimate5)
xgb_classifier5.fit(X_train5, y_train5)
xgb.plot_importance(xgb_classifier5)
plt.show()
# training set
y_pred_xgb_tr5 = xgb_classifier5.predict(X_train5)
accuracy_xgb_tr5 = accuracy_score(y_train5, y_pred_xgb_tr5)
print("XGBoost Model Score (Training Set): %.2f%%" % (accuracy_xgb_tr5 * 100.0))
# testing set
y_pred_xgb_t5 = xgb_classifier5.predict(X_test5)
xgb_predictions5 = [round(value) for value in y_pred_xgb_t5]
# evaluate predictions
accuracy_xgb_t5 = accuracy_score(y_test5, xgb_predictions5)
print("XGBoost Model Score (Testing Set): %.2f%%" % (accuracy_xgb_t5 * 100.0))
XGBoost Model Score (Training Set): 100.00% XGBoost Model Score (Testing Set): 72.84%
plot_confusion_matrix(xgb_classifier5, X_test5, y_test5,cmap="YlGnBu"); # good improvement
# Printing the classification report
print(classification_report(y_test5, y_pred_xgb_t5))
precision recall f1-score support
0 0.78 0.81 0.79 318
1 0.63 0.59 0.61 179
accuracy 0.73 497
macro avg 0.70 0.70 0.70 497
weighted avg 0.72 0.73 0.73 497
# Accuracy Score for each company.
xgb_accuracy = {'Aramco': accuracy_xgb_t1*100, 'SARCO': accuracy_xgb_t2*100, 'Rabigh': accuracy_xgb_t3*100, 'Bahri':accuracy_xgb_t4*100, 'Aldrees': accuracy_xgb_t5*100}
print(xgb_accuracy)
xgb_names = list(xgb_accuracy.keys())
xgb_values = list(xgb_accuracy.values())
fig, ax = plt.subplots(figsize=(15,8))
plt.title('XGBoost Accuracy Score per Company')
bars = ax.barh(xgb_names, xgb_values)
ax.bar_label(bars, fmt='%.2f%%')
plt.show()
{'Aramco': 66.4179104477612, 'SARCO': 93.56136820925553, 'Rabigh': 93.96378269617706, 'Bahri': 64.98993963782696, 'Aldrees': 72.83702213279678}
#!python --version # update to latest version
# https://stackoverflow.com/questions/68657341/how-can-i-update-google-colabs-python-version#:~:text=Each%20runtime%20gives%20you%20a%20fresh%20google%20colab.&text=python%20version%20manually)-,!,and%20update%20the%20default%20python.
# Define the search space
xgboost = xgb.XGBClassifier(objective= "binary:logistic", random_state=42)
param_grid = {
# Learning rate shrinks the weights to make the boosting process more conservative
"learning_rate": [0.0001,0.001, 0.01, 0.02, 0.03, 0.04, 0.05, 0.1] ,
# Percentage of columns to be randomly samples for each tree.
'colsample_bytree': [0.3, 0.7, 1, 2, 3],
'base_score':[0.5, 0.7],
# Maximum depth of the tree, increasing it increases the model complexity.
'max_depth': [2, 4, 5, 6, 8, 10, 12],
'subsample': [0.2, 0.3, 0.5, 0.7, 0.8, 1],
'n_estimators': [100],
'scale_pos_weight': [estimate1]}
# Set up score
scoring = ['accuracy']
# Set up the k-fold cross-validation
kfold = KFold(n_splits=5, shuffle=True, random_state=42)
# Define random search
xgb_random_search = RandomizedSearchCV(estimator=xgboost,
param_distributions=param_grid,
n_iter=32,
scoring=scoring,
n_jobs=-1,
cv=kfold,
refit='accuracy',
verbose=0)
xgb_random_search
RandomizedSearchCV(cv=KFold(n_splits=5, random_state=42, shuffle=True),
estimator=XGBClassifier(base_score=None, booster=None,
callbacks=None,
colsample_bylevel=None,
colsample_bynode=None,
colsample_bytree=None,
early_stopping_rounds=None,
enable_categorical=False,
eval_metric=None, gamma=None,
gpu_id=None, grow_policy=None,
importance_type=None,
interaction_cons...
reg_alpha=None, reg_lambda=None, ...),
n_iter=32, n_jobs=-1,
param_distributions={'base_score': [0.5, 0.7],
'colsample_bytree': [0.3, 0.7, 1, 2, 3],
'learning_rate': [0.0001, 0.001, 0.01,
0.02, 0.03, 0.04,
0.05, 0.1],
'max_depth': [2, 4, 5, 6, 8, 10, 12],
'n_estimators': [100],
'scale_pos_weight': [0.7631578947368421],
'subsample': [0.2, 0.3, 0.5, 0.7, 0.8,
1]},
refit='accuracy', scoring=['accuracy'])In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. RandomizedSearchCV(cv=KFold(n_splits=5, random_state=42, shuffle=True),
estimator=XGBClassifier(base_score=None, booster=None,
callbacks=None,
colsample_bylevel=None,
colsample_bynode=None,
colsample_bytree=None,
early_stopping_rounds=None,
enable_categorical=False,
eval_metric=None, gamma=None,
gpu_id=None, grow_policy=None,
importance_type=None,
interaction_cons...
reg_alpha=None, reg_lambda=None, ...),
n_iter=32, n_jobs=-1,
param_distributions={'base_score': [0.5, 0.7],
'colsample_bytree': [0.3, 0.7, 1, 2, 3],
'learning_rate': [0.0001, 0.001, 0.01,
0.02, 0.03, 0.04,
0.05, 0.1],
'max_depth': [2, 4, 5, 6, 8, 10, 12],
'n_estimators': [100],
'scale_pos_weight': [0.7631578947368421],
'subsample': [0.2, 0.3, 0.5, 0.7, 0.8,
1]},
refit='accuracy', scoring=['accuracy'])XGBClassifier(base_score=None, booster=None, callbacks=None,
colsample_bylevel=None, colsample_bynode=None,
colsample_bytree=None, early_stopping_rounds=None,
enable_categorical=False, eval_metric=None, gamma=None,
gpu_id=None, grow_policy=None, importance_type=None,
interaction_constraints=None, learning_rate=None, max_bin=None,
max_cat_to_onehot=None, max_delta_step=None, max_depth=None,
max_leaves=None, min_child_weight=None, missing=nan,
monotone_constraints=None, n_estimators=100, n_jobs=None,
num_parallel_tree=None, predictor=None, random_state=42,
reg_alpha=None, reg_lambda=None, ...)XGBClassifier(base_score=None, booster=None, callbacks=None,
colsample_bylevel=None, colsample_bynode=None,
colsample_bytree=None, early_stopping_rounds=None,
enable_categorical=False, eval_metric=None, gamma=None,
gpu_id=None, grow_policy=None, importance_type=None,
interaction_constraints=None, learning_rate=None, max_bin=None,
max_cat_to_onehot=None, max_delta_step=None, max_depth=None,
max_leaves=None, min_child_weight=None, missing=nan,
monotone_constraints=None, n_estimators=100, n_jobs=None,
num_parallel_tree=None, predictor=None, random_state=42,
reg_alpha=None, reg_lambda=None, ...)# Fit grid search
# take only relevant columns to improve accuracy.
X_train11, X_test11, y_train11, y_test11 = split_data(df_Aramco.drop(['EMA_9'], axis=1), 'Target')
random_result1 = xgb_random_search.fit(X_train11, y_train11)
# Print grid search summary
random_result1
RandomizedSearchCV(cv=KFold(n_splits=5, random_state=42, shuffle=True),
estimator=XGBClassifier(base_score=None, booster=None,
callbacks=None,
colsample_bylevel=None,
colsample_bynode=None,
colsample_bytree=None,
early_stopping_rounds=None,
enable_categorical=False,
eval_metric=None, gamma=None,
gpu_id=None, grow_policy=None,
importance_type=None,
interaction_cons...
reg_alpha=None, reg_lambda=None, ...),
n_iter=32, n_jobs=-1,
param_distributions={'base_score': [0.5, 0.7],
'colsample_bytree': [0.3, 0.7, 1, 2, 3],
'learning_rate': [0.0001, 0.001, 0.01,
0.02, 0.03, 0.04,
0.05, 0.1],
'max_depth': [2, 4, 5, 6, 8, 10, 12],
'n_estimators': [100],
'scale_pos_weight': [0.7631578947368421],
'subsample': [0.2, 0.3, 0.5, 0.7, 0.8,
1]},
refit='accuracy', scoring=['accuracy'])In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. RandomizedSearchCV(cv=KFold(n_splits=5, random_state=42, shuffle=True),
estimator=XGBClassifier(base_score=None, booster=None,
callbacks=None,
colsample_bylevel=None,
colsample_bynode=None,
colsample_bytree=None,
early_stopping_rounds=None,
enable_categorical=False,
eval_metric=None, gamma=None,
gpu_id=None, grow_policy=None,
importance_type=None,
interaction_cons...
reg_alpha=None, reg_lambda=None, ...),
n_iter=32, n_jobs=-1,
param_distributions={'base_score': [0.5, 0.7],
'colsample_bytree': [0.3, 0.7, 1, 2, 3],
'learning_rate': [0.0001, 0.001, 0.01,
0.02, 0.03, 0.04,
0.05, 0.1],
'max_depth': [2, 4, 5, 6, 8, 10, 12],
'n_estimators': [100],
'scale_pos_weight': [0.7631578947368421],
'subsample': [0.2, 0.3, 0.5, 0.7, 0.8,
1]},
refit='accuracy', scoring=['accuracy'])XGBClassifier(base_score=None, booster=None, callbacks=None,
colsample_bylevel=None, colsample_bynode=None,
colsample_bytree=None, early_stopping_rounds=None,
enable_categorical=False, eval_metric=None, gamma=None,
gpu_id=None, grow_policy=None, importance_type=None,
interaction_constraints=None, learning_rate=None, max_bin=None,
max_cat_to_onehot=None, max_delta_step=None, max_depth=None,
max_leaves=None, min_child_weight=None, missing=nan,
monotone_constraints=None, n_estimators=100, n_jobs=None,
num_parallel_tree=None, predictor=None, random_state=42,
reg_alpha=None, reg_lambda=None, ...)XGBClassifier(base_score=None, booster=None, callbacks=None,
colsample_bylevel=None, colsample_bynode=None,
colsample_bytree=None, early_stopping_rounds=None,
enable_categorical=False, eval_metric=None, gamma=None,
gpu_id=None, grow_policy=None, importance_type=None,
interaction_constraints=None, learning_rate=None, max_bin=None,
max_cat_to_onehot=None, max_delta_step=None, max_depth=None,
max_leaves=None, min_child_weight=None, missing=nan,
monotone_constraints=None, n_estimators=100, n_jobs=None,
num_parallel_tree=None, predictor=None, random_state=42,
reg_alpha=None, reg_lambda=None, ...)print("\n\nBest score: %f using %s" % (random_result1.best_score_, random_result1.best_params_))
print("Best Estimator: ")
xgb_random_search.best_estimator_
Best score: 0.621184 using {'subsample': 0.2, 'scale_pos_weight': 0.7631578947368421, 'n_estimators': 100, 'max_depth': 2, 'learning_rate': 0.0001, 'colsample_bytree': 1, 'base_score': 0.5}
Best Estimator:
XGBClassifier(base_score=0.5, booster='gbtree', callbacks=None,
colsample_bylevel=1, colsample_bynode=1, colsample_bytree=1,
early_stopping_rounds=None, enable_categorical=False,
eval_metric=None, gamma=0, gpu_id=-1, grow_policy='depthwise',
importance_type=None, interaction_constraints='',
learning_rate=0.0001, max_bin=256, max_cat_to_onehot=4,
max_delta_step=0, max_depth=2, max_leaves=0, min_child_weight=1,
missing=nan, monotone_constraints='()', n_estimators=100,
n_jobs=0, num_parallel_tree=1, predictor='auto', random_state=42,
reg_alpha=0, reg_lambda=1, ...)In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. XGBClassifier(base_score=0.5, booster='gbtree', callbacks=None,
colsample_bylevel=1, colsample_bynode=1, colsample_bytree=1,
early_stopping_rounds=None, enable_categorical=False,
eval_metric=None, gamma=0, gpu_id=-1, grow_policy='depthwise',
importance_type=None, interaction_constraints='',
learning_rate=0.0001, max_bin=256, max_cat_to_onehot=4,
max_delta_step=0, max_depth=2, max_leaves=0, min_child_weight=1,
missing=nan, monotone_constraints='()', n_estimators=100,
n_jobs=0, num_parallel_tree=1, predictor='auto', random_state=42,
reg_alpha=0, reg_lambda=1, ...)# training set
y_pred_xgb_tr111 = xgb_random_search.predict(X_train11)
accuracy_xgb_tr111 = accuracy_score(y_train11, y_pred_xgb_tr111)
print("XGBoost Model Score (Training Set): %.2f%%" % (accuracy_xgb_tr111 * 100.0))
# testing set
y_pred_xgb_t111 = xgb_random_search.predict(X_test11)
xgb_predictions111 = [round(value) for value in y_pred_xgb_t111]
# evaluate predictions
accuracy_xgb_t111 = accuracy_score(y_test11, xgb_predictions111)
print("XGBoost Model Score (Testing Set): %.2f%%" % (accuracy_xgb_t111 * 100.0))
XGBoost Model Score (Training Set): 67.16% XGBoost Model Score (Testing Set): 67.91%
plot_confusion_matrix(xgb_random_search, X_test11, y_test11,cmap="YlGnBu"); # good improvement
# Printing the classification report
print(classification_report(y_test11, y_pred_xgb_t111))
precision recall f1-score support
0 0.58 0.76 0.66 54
1 0.79 0.62 0.70 80
accuracy 0.68 134
macro avg 0.69 0.69 0.68 134
weighted avg 0.71 0.68 0.68 134
X_train22, X_test22, y_train22, y_test22 = split_data(df_SARCO.drop(['EMA_9'], axis=1), 'Target')
X_train22, X_test22 = scale_data(X_train22, X_test22)
# fit random search
param_grid['scale_pos_weight'] = [estimate2]
xgb_random_search2 = RandomizedSearchCV(estimator=xgboost,
param_distributions=param_grid,
n_iter=32,
scoring=scoring,
n_jobs=-1,
cv=kfold,
refit='accuracy',
verbose=0)
random_result2 = xgb_random_search2.fit(X_train22, y_train22)
random_result2
RandomizedSearchCV(cv=KFold(n_splits=5, random_state=42, shuffle=True),
estimator=XGBClassifier(base_score=None, booster=None,
callbacks=None,
colsample_bylevel=None,
colsample_bynode=None,
colsample_bytree=None,
early_stopping_rounds=None,
enable_categorical=False,
eval_metric=None, gamma=None,
gpu_id=None, grow_policy=None,
importance_type=None,
interaction_cons...
reg_alpha=None, reg_lambda=None, ...),
n_iter=32, n_jobs=-1,
param_distributions={'base_score': [0.5, 0.7],
'colsample_bytree': [0.3, 0.7, 1, 2, 3],
'learning_rate': [0.0001, 0.001, 0.01,
0.02, 0.03, 0.04,
0.05, 0.1],
'max_depth': [2, 4, 5, 6, 8, 10, 12],
'n_estimators': [100],
'scale_pos_weight': [16.110344827586207],
'subsample': [0.2, 0.3, 0.5, 0.7, 0.8,
1]},
refit='accuracy', scoring=['accuracy'])In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. RandomizedSearchCV(cv=KFold(n_splits=5, random_state=42, shuffle=True),
estimator=XGBClassifier(base_score=None, booster=None,
callbacks=None,
colsample_bylevel=None,
colsample_bynode=None,
colsample_bytree=None,
early_stopping_rounds=None,
enable_categorical=False,
eval_metric=None, gamma=None,
gpu_id=None, grow_policy=None,
importance_type=None,
interaction_cons...
reg_alpha=None, reg_lambda=None, ...),
n_iter=32, n_jobs=-1,
param_distributions={'base_score': [0.5, 0.7],
'colsample_bytree': [0.3, 0.7, 1, 2, 3],
'learning_rate': [0.0001, 0.001, 0.01,
0.02, 0.03, 0.04,
0.05, 0.1],
'max_depth': [2, 4, 5, 6, 8, 10, 12],
'n_estimators': [100],
'scale_pos_weight': [16.110344827586207],
'subsample': [0.2, 0.3, 0.5, 0.7, 0.8,
1]},
refit='accuracy', scoring=['accuracy'])XGBClassifier(base_score=None, booster=None, callbacks=None,
colsample_bylevel=None, colsample_bynode=None,
colsample_bytree=None, early_stopping_rounds=None,
enable_categorical=False, eval_metric=None, gamma=None,
gpu_id=None, grow_policy=None, importance_type=None,
interaction_constraints=None, learning_rate=None, max_bin=None,
max_cat_to_onehot=None, max_delta_step=None, max_depth=None,
max_leaves=None, min_child_weight=None, missing=nan,
monotone_constraints=None, n_estimators=100, n_jobs=None,
num_parallel_tree=None, predictor=None, random_state=42,
reg_alpha=None, reg_lambda=None, ...)XGBClassifier(base_score=None, booster=None, callbacks=None,
colsample_bylevel=None, colsample_bynode=None,
colsample_bytree=None, early_stopping_rounds=None,
enable_categorical=False, eval_metric=None, gamma=None,
gpu_id=None, grow_policy=None, importance_type=None,
interaction_constraints=None, learning_rate=None, max_bin=None,
max_cat_to_onehot=None, max_delta_step=None, max_depth=None,
max_leaves=None, min_child_weight=None, missing=nan,
monotone_constraints=None, n_estimators=100, n_jobs=None,
num_parallel_tree=None, predictor=None, random_state=42,
reg_alpha=None, reg_lambda=None, ...)print("\n\nBest score: %f using %s" % (random_result2.best_score_, random_result2.best_params_))
print("Best Estimator: ")
xgb_random_search2.best_estimator_
Best score: 0.931953 using {'subsample': 0.5, 'scale_pos_weight': 16.110344827586207, 'n_estimators': 100, 'max_depth': 12, 'learning_rate': 0.03, 'colsample_bytree': 0.3, 'base_score': 0.7}
Best Estimator:
XGBClassifier(base_score=0.7, booster='gbtree', callbacks=None,
colsample_bylevel=1, colsample_bynode=1, colsample_bytree=0.3,
early_stopping_rounds=None, enable_categorical=False,
eval_metric=None, gamma=0, gpu_id=-1, grow_policy='depthwise',
importance_type=None, interaction_constraints='',
learning_rate=0.03, max_bin=256, max_cat_to_onehot=4,
max_delta_step=0, max_depth=12, max_leaves=0, min_child_weight=1,
missing=nan, monotone_constraints='()', n_estimators=100,
n_jobs=0, num_parallel_tree=1, predictor='auto', random_state=42,
reg_alpha=0, reg_lambda=1, ...)In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. XGBClassifier(base_score=0.7, booster='gbtree', callbacks=None,
colsample_bylevel=1, colsample_bynode=1, colsample_bytree=0.3,
early_stopping_rounds=None, enable_categorical=False,
eval_metric=None, gamma=0, gpu_id=-1, grow_policy='depthwise',
importance_type=None, interaction_constraints='',
learning_rate=0.03, max_bin=256, max_cat_to_onehot=4,
max_delta_step=0, max_depth=12, max_leaves=0, min_child_weight=1,
missing=nan, monotone_constraints='()', n_estimators=100,
n_jobs=0, num_parallel_tree=1, predictor='auto', random_state=42,
reg_alpha=0, reg_lambda=1, ...)# training set
y_pred_xgb_tr222 = xgb_random_search2.predict(X_train22)
accuracy_xgb_tr222 = accuracy_score(y_train22, y_pred_xgb_tr222)
print("XGBoost Model Score (Training Set): %.2f%%" % (accuracy_xgb_tr222 * 100.0))
# testing set
y_pred_xgb_t222 = xgb_random_search2.predict(X_test22)
xgb_predictions222 = [round(value) for value in y_pred_xgb_t222]
# evaluate predictions
accuracy_xgb_t222 = accuracy_score(y_test22, xgb_predictions222)
print("XGBoost Model Score (Testing Set): %.2f%%" % (accuracy_xgb_t222 * 100.0))
XGBoost Model Score (Training Set): 99.70% XGBoost Model Score (Testing Set): 94.57%
plot_confusion_matrix(xgb_random_search2, X_test22, y_test22,cmap="YlGnBu"); # good improvement
# Printing the classification report
print(classification_report(y_test22, y_pred_xgb_t222))
precision recall f1-score support
0 0.96 0.98 0.97 468
1 0.55 0.41 0.47 29
accuracy 0.95 497
macro avg 0.75 0.70 0.72 497
weighted avg 0.94 0.95 0.94 497
X_train33, X_test33, y_train33, y_test33 = split_data(df_Rabigh.drop(['EMA_9'], axis=1), 'Target')
X_train33, X_test33 = scale_data(X_train33, X_test33)
param_grid['scale_pos_weight'] = [estimate3]
xgb_random_search3 = RandomizedSearchCV(estimator=xgboost,
param_distributions=param_grid,
n_iter=32,
scoring=scoring,
n_jobs=-1,
cv=kfold,
refit='accuracy',
verbose=0)
random_result3 = xgb_random_search3.fit(X_train33, y_train33)
random_result3
RandomizedSearchCV(cv=KFold(n_splits=5, random_state=42, shuffle=True),
estimator=XGBClassifier(base_score=None, booster=None,
callbacks=None,
colsample_bylevel=None,
colsample_bynode=None,
colsample_bytree=None,
early_stopping_rounds=None,
enable_categorical=False,
eval_metric=None, gamma=None,
gpu_id=None, grow_policy=None,
importance_type=None,
interaction_cons...
reg_alpha=None, reg_lambda=None, ...),
n_iter=32, n_jobs=-1,
param_distributions={'base_score': [0.5, 0.7],
'colsample_bytree': [0.3, 0.7, 1, 2, 3],
'learning_rate': [0.0001, 0.001, 0.01,
0.02, 0.03, 0.04,
0.05, 0.1],
'max_depth': [2, 4, 5, 6, 8, 10, 12],
'n_estimators': [100],
'scale_pos_weight': [0.0324459234608985],
'subsample': [0.2, 0.3, 0.5, 0.7, 0.8,
1]},
refit='accuracy', scoring=['accuracy'])In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. RandomizedSearchCV(cv=KFold(n_splits=5, random_state=42, shuffle=True),
estimator=XGBClassifier(base_score=None, booster=None,
callbacks=None,
colsample_bylevel=None,
colsample_bynode=None,
colsample_bytree=None,
early_stopping_rounds=None,
enable_categorical=False,
eval_metric=None, gamma=None,
gpu_id=None, grow_policy=None,
importance_type=None,
interaction_cons...
reg_alpha=None, reg_lambda=None, ...),
n_iter=32, n_jobs=-1,
param_distributions={'base_score': [0.5, 0.7],
'colsample_bytree': [0.3, 0.7, 1, 2, 3],
'learning_rate': [0.0001, 0.001, 0.01,
0.02, 0.03, 0.04,
0.05, 0.1],
'max_depth': [2, 4, 5, 6, 8, 10, 12],
'n_estimators': [100],
'scale_pos_weight': [0.0324459234608985],
'subsample': [0.2, 0.3, 0.5, 0.7, 0.8,
1]},
refit='accuracy', scoring=['accuracy'])XGBClassifier(base_score=None, booster=None, callbacks=None,
colsample_bylevel=None, colsample_bynode=None,
colsample_bytree=None, early_stopping_rounds=None,
enable_categorical=False, eval_metric=None, gamma=None,
gpu_id=None, grow_policy=None, importance_type=None,
interaction_constraints=None, learning_rate=None, max_bin=None,
max_cat_to_onehot=None, max_delta_step=None, max_depth=None,
max_leaves=None, min_child_weight=None, missing=nan,
monotone_constraints=None, n_estimators=100, n_jobs=None,
num_parallel_tree=None, predictor=None, random_state=42,
reg_alpha=None, reg_lambda=None, ...)XGBClassifier(base_score=None, booster=None, callbacks=None,
colsample_bylevel=None, colsample_bynode=None,
colsample_bytree=None, early_stopping_rounds=None,
enable_categorical=False, eval_metric=None, gamma=None,
gpu_id=None, grow_policy=None, importance_type=None,
interaction_constraints=None, learning_rate=None, max_bin=None,
max_cat_to_onehot=None, max_delta_step=None, max_depth=None,
max_leaves=None, min_child_weight=None, missing=nan,
monotone_constraints=None, n_estimators=100, n_jobs=None,
num_parallel_tree=None, predictor=None, random_state=42,
reg_alpha=None, reg_lambda=None, ...)print("\n\nBest score: %f using %s" % (random_result3.best_score_, random_result3.best_params_))
print("Best Estimator: ")
xgb_random_search3.best_estimator_
Best score: 0.969270 using {'subsample': 0.2, 'scale_pos_weight': 0.0324459234608985, 'n_estimators': 100, 'max_depth': 6, 'learning_rate': 0.001, 'colsample_bytree': 0.3, 'base_score': 0.7}
Best Estimator:
XGBClassifier(base_score=0.7, booster='gbtree', callbacks=None,
colsample_bylevel=1, colsample_bynode=1, colsample_bytree=0.3,
early_stopping_rounds=None, enable_categorical=False,
eval_metric=None, gamma=0, gpu_id=-1, grow_policy='depthwise',
importance_type=None, interaction_constraints='',
learning_rate=0.001, max_bin=256, max_cat_to_onehot=4,
max_delta_step=0, max_depth=6, max_leaves=0, min_child_weight=1,
missing=nan, monotone_constraints='()', n_estimators=100,
n_jobs=0, num_parallel_tree=1, predictor='auto', random_state=42,
reg_alpha=0, reg_lambda=1, ...)In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. XGBClassifier(base_score=0.7, booster='gbtree', callbacks=None,
colsample_bylevel=1, colsample_bynode=1, colsample_bytree=0.3,
early_stopping_rounds=None, enable_categorical=False,
eval_metric=None, gamma=0, gpu_id=-1, grow_policy='depthwise',
importance_type=None, interaction_constraints='',
learning_rate=0.001, max_bin=256, max_cat_to_onehot=4,
max_delta_step=0, max_depth=6, max_leaves=0, min_child_weight=1,
missing=nan, monotone_constraints='()', n_estimators=100,
n_jobs=0, num_parallel_tree=1, predictor='auto', random_state=42,
reg_alpha=0, reg_lambda=1, ...)# training set
y_pred_xgb_tr333 = xgb_random_search3.predict(X_train33)
accuracy_xgb_tr333 = accuracy_score(y_train33, y_pred_xgb_tr333)
print("XGBoost Model Score (Training Set): %.2f%%" % (accuracy_xgb_tr333 * 100.0))
# testing set
y_pred_xgb_t333 = xgb_random_search3.predict(X_test33)
xgb_predictions333 = [round(value) for value in y_pred_xgb_t333]
# evaluate predictions
accuracy_xgb_t333 = accuracy_score(y_test33, xgb_predictions333)
print("XGBoost Model Score (Testing Set): %.2f%%" % (accuracy_xgb_t333 * 100.0))
XGBoost Model Score (Training Set): 96.93% XGBoost Model Score (Testing Set): 96.58%
plot_confusion_matrix(xgb_random_search3, X_test33, y_test33,cmap="YlGnBu"); # good improvement
# Printing the classification report
print(classification_report(y_test33, y_pred_xgb_t333))
precision recall f1-score support
0 0.00 0.00 0.00 17
1 0.97 1.00 0.98 480
accuracy 0.97 497
macro avg 0.48 0.50 0.49 497
weighted avg 0.93 0.97 0.95 497
X_train44, X_test44, y_train44, y_test44 = split_data(df_Bahri.drop(['EMA_9'], axis=1), 'Target')
X_train44, X_test44 = scale_data(X_train44, X_test44)
param_grid['scale_pos_weight'] = [estimate4]
xgb_random_search4 = RandomizedSearchCV(estimator=xgboost,
param_distributions=param_grid,
n_iter=32,
scoring=scoring,
n_jobs=-1,
cv=kfold,
refit='accuracy',
verbose=0)
random_result4 = xgb_random_search4.fit(X_train44, y_train44)
random_result4
RandomizedSearchCV(cv=KFold(n_splits=5, random_state=42, shuffle=True),
estimator=XGBClassifier(base_score=None, booster=None,
callbacks=None,
colsample_bylevel=None,
colsample_bynode=None,
colsample_bytree=None,
early_stopping_rounds=None,
enable_categorical=False,
eval_metric=None, gamma=None,
gpu_id=None, grow_policy=None,
importance_type=None,
interaction_cons...
reg_alpha=None, reg_lambda=None, ...),
n_iter=32, n_jobs=-1,
param_distributions={'base_score': [0.5, 0.7],
'colsample_bytree': [0.3, 0.7, 1, 2, 3],
'learning_rate': [0.0001, 0.001, 0.01,
0.02, 0.03, 0.04,
0.05, 0.1],
'max_depth': [2, 4, 5, 6, 8, 10, 12],
'n_estimators': [100],
'scale_pos_weight': [0.7508821453775583],
'subsample': [0.2, 0.3, 0.5, 0.7, 0.8,
1]},
refit='accuracy', scoring=['accuracy'])In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. RandomizedSearchCV(cv=KFold(n_splits=5, random_state=42, shuffle=True),
estimator=XGBClassifier(base_score=None, booster=None,
callbacks=None,
colsample_bylevel=None,
colsample_bynode=None,
colsample_bytree=None,
early_stopping_rounds=None,
enable_categorical=False,
eval_metric=None, gamma=None,
gpu_id=None, grow_policy=None,
importance_type=None,
interaction_cons...
reg_alpha=None, reg_lambda=None, ...),
n_iter=32, n_jobs=-1,
param_distributions={'base_score': [0.5, 0.7],
'colsample_bytree': [0.3, 0.7, 1, 2, 3],
'learning_rate': [0.0001, 0.001, 0.01,
0.02, 0.03, 0.04,
0.05, 0.1],
'max_depth': [2, 4, 5, 6, 8, 10, 12],
'n_estimators': [100],
'scale_pos_weight': [0.7508821453775583],
'subsample': [0.2, 0.3, 0.5, 0.7, 0.8,
1]},
refit='accuracy', scoring=['accuracy'])XGBClassifier(base_score=None, booster=None, callbacks=None,
colsample_bylevel=None, colsample_bynode=None,
colsample_bytree=None, early_stopping_rounds=None,
enable_categorical=False, eval_metric=None, gamma=None,
gpu_id=None, grow_policy=None, importance_type=None,
interaction_constraints=None, learning_rate=None, max_bin=None,
max_cat_to_onehot=None, max_delta_step=None, max_depth=None,
max_leaves=None, min_child_weight=None, missing=nan,
monotone_constraints=None, n_estimators=100, n_jobs=None,
num_parallel_tree=None, predictor=None, random_state=42,
reg_alpha=None, reg_lambda=None, ...)XGBClassifier(base_score=None, booster=None, callbacks=None,
colsample_bylevel=None, colsample_bynode=None,
colsample_bytree=None, early_stopping_rounds=None,
enable_categorical=False, eval_metric=None, gamma=None,
gpu_id=None, grow_policy=None, importance_type=None,
interaction_constraints=None, learning_rate=None, max_bin=None,
max_cat_to_onehot=None, max_delta_step=None, max_depth=None,
max_leaves=None, min_child_weight=None, missing=nan,
monotone_constraints=None, n_estimators=100, n_jobs=None,
num_parallel_tree=None, predictor=None, random_state=42,
reg_alpha=None, reg_lambda=None, ...)print("\n\nBest score: %f using %s" % (random_result4.best_score_, random_result4.best_params_))
print("Best Estimator: ")
xgb_random_search4.best_estimator_
Best score: 0.695555 using {'subsample': 0.5, 'scale_pos_weight': 0.7508821453775583, 'n_estimators': 100, 'max_depth': 2, 'learning_rate': 0.03, 'colsample_bytree': 0.7, 'base_score': 0.7}
Best Estimator:
XGBClassifier(base_score=0.7, booster='gbtree', callbacks=None,
colsample_bylevel=1, colsample_bynode=1, colsample_bytree=0.7,
early_stopping_rounds=None, enable_categorical=False,
eval_metric=None, gamma=0, gpu_id=-1, grow_policy='depthwise',
importance_type=None, interaction_constraints='',
learning_rate=0.03, max_bin=256, max_cat_to_onehot=4,
max_delta_step=0, max_depth=2, max_leaves=0, min_child_weight=1,
missing=nan, monotone_constraints='()', n_estimators=100,
n_jobs=0, num_parallel_tree=1, predictor='auto', random_state=42,
reg_alpha=0, reg_lambda=1, ...)In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. XGBClassifier(base_score=0.7, booster='gbtree', callbacks=None,
colsample_bylevel=1, colsample_bynode=1, colsample_bytree=0.7,
early_stopping_rounds=None, enable_categorical=False,
eval_metric=None, gamma=0, gpu_id=-1, grow_policy='depthwise',
importance_type=None, interaction_constraints='',
learning_rate=0.03, max_bin=256, max_cat_to_onehot=4,
max_delta_step=0, max_depth=2, max_leaves=0, min_child_weight=1,
missing=nan, monotone_constraints='()', n_estimators=100,
n_jobs=0, num_parallel_tree=1, predictor='auto', random_state=42,
reg_alpha=0, reg_lambda=1, ...)# xgb_classifier4 = XGBClassifier(base_score=0.7, colsample_bytree=0.7,
# learning_rate=0.1,max_depth=8, n_estimators=100,
# random_state=42, subsample=0.5)
# training set
#xgb_class4 = xgb_random_search.fit(X_train4, y_train4)# training set
y_pred_xgb_tr444 = xgb_random_search4.predict(X_train44)
accuracy_xgb_tr444 = accuracy_score(y_train44, y_pred_xgb_tr444)
print("XGBoost Model Score (Training Set): %.2f%%" % (accuracy_xgb_tr444 * 100.0))
# testing set
y_pred_xgb_t444 = xgb_random_search4.predict(X_test44)
xgb_predictions444 = [round(value) for value in y_pred_xgb_t444]
# evaluate predictions
accuracy_xgb_t444 = accuracy_score(y_test44, xgb_predictions444)
print("XGBoost Model Score (Testing Set): %.2f%%" % (accuracy_xgb_t444 * 100.0))
XGBoost Model Score (Training Set): 71.02% XGBoost Model Score (Testing Set): 67.20%
plot_confusion_matrix(xgb_random_search4, X_test44, y_test44,cmap="YlGnBu"); # good improvement
# Printing the classification report
print(classification_report(y_test44, y_pred_xgb_t444))
precision recall f1-score support
0 0.63 0.64 0.63 222
1 0.70 0.70 0.70 275
accuracy 0.67 497
macro avg 0.67 0.67 0.67 497
weighted avg 0.67 0.67 0.67 497
X_train55, X_test55, y_train55, y_test55 = split_data(df_Aldrees.drop(['EMA_9'], axis=1), 'Target')
X_train55, X_test55 = scale_data(X_train55, X_test55)
param_grid['scale_pos_weight'] = [estimate5]
xgb_random_search5 = RandomizedSearchCV(estimator=xgboost,
param_distributions=param_grid,
n_iter=32,
scoring=scoring,
n_jobs=-1,
cv=kfold,
refit='accuracy',
verbose=0)
random_result5 = xgb_random_search5.fit(X_train55, y_train55)
random_result5
RandomizedSearchCV(cv=KFold(n_splits=5, random_state=42, shuffle=True),
estimator=XGBClassifier(base_score=None, booster=None,
callbacks=None,
colsample_bylevel=None,
colsample_bynode=None,
colsample_bytree=None,
early_stopping_rounds=None,
enable_categorical=False,
eval_metric=None, gamma=None,
gpu_id=None, grow_policy=None,
importance_type=None,
interaction_cons...
reg_alpha=None, reg_lambda=None, ...),
n_iter=32, n_jobs=-1,
param_distributions={'base_score': [0.5, 0.7],
'colsample_bytree': [0.3, 0.7, 1, 2, 3],
'learning_rate': [0.0001, 0.001, 0.01,
0.02, 0.03, 0.04,
0.05, 0.1],
'max_depth': [2, 4, 5, 6, 8, 10, 12],
'n_estimators': [100],
'scale_pos_weight': [1.7486157253599115],
'subsample': [0.2, 0.3, 0.5, 0.7, 0.8,
1]},
refit='accuracy', scoring=['accuracy'])In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. RandomizedSearchCV(cv=KFold(n_splits=5, random_state=42, shuffle=True),
estimator=XGBClassifier(base_score=None, booster=None,
callbacks=None,
colsample_bylevel=None,
colsample_bynode=None,
colsample_bytree=None,
early_stopping_rounds=None,
enable_categorical=False,
eval_metric=None, gamma=None,
gpu_id=None, grow_policy=None,
importance_type=None,
interaction_cons...
reg_alpha=None, reg_lambda=None, ...),
n_iter=32, n_jobs=-1,
param_distributions={'base_score': [0.5, 0.7],
'colsample_bytree': [0.3, 0.7, 1, 2, 3],
'learning_rate': [0.0001, 0.001, 0.01,
0.02, 0.03, 0.04,
0.05, 0.1],
'max_depth': [2, 4, 5, 6, 8, 10, 12],
'n_estimators': [100],
'scale_pos_weight': [1.7486157253599115],
'subsample': [0.2, 0.3, 0.5, 0.7, 0.8,
1]},
refit='accuracy', scoring=['accuracy'])XGBClassifier(base_score=None, booster=None, callbacks=None,
colsample_bylevel=None, colsample_bynode=None,
colsample_bytree=None, early_stopping_rounds=None,
enable_categorical=False, eval_metric=None, gamma=None,
gpu_id=None, grow_policy=None, importance_type=None,
interaction_constraints=None, learning_rate=None, max_bin=None,
max_cat_to_onehot=None, max_delta_step=None, max_depth=None,
max_leaves=None, min_child_weight=None, missing=nan,
monotone_constraints=None, n_estimators=100, n_jobs=None,
num_parallel_tree=None, predictor=None, random_state=42,
reg_alpha=None, reg_lambda=None, ...)XGBClassifier(base_score=None, booster=None, callbacks=None,
colsample_bylevel=None, colsample_bynode=None,
colsample_bytree=None, early_stopping_rounds=None,
enable_categorical=False, eval_metric=None, gamma=None,
gpu_id=None, grow_policy=None, importance_type=None,
interaction_constraints=None, learning_rate=None, max_bin=None,
max_cat_to_onehot=None, max_delta_step=None, max_depth=None,
max_leaves=None, min_child_weight=None, missing=nan,
monotone_constraints=None, n_estimators=100, n_jobs=None,
num_parallel_tree=None, predictor=None, random_state=42,
reg_alpha=None, reg_lambda=None, ...)print("\n\nBest score: %f using %s" % (random_result5.best_score_, random_result5.best_params_))
print("Best Estimator: ")
xgb_random_search5.best_estimator_
Best score: 0.703275 using {'subsample': 0.3, 'scale_pos_weight': 1.7486157253599115, 'n_estimators': 100, 'max_depth': 8, 'learning_rate': 0.001, 'colsample_bytree': 1, 'base_score': 0.5}
Best Estimator:
XGBClassifier(base_score=0.5, booster='gbtree', callbacks=None,
colsample_bylevel=1, colsample_bynode=1, colsample_bytree=1,
early_stopping_rounds=None, enable_categorical=False,
eval_metric=None, gamma=0, gpu_id=-1, grow_policy='depthwise',
importance_type=None, interaction_constraints='',
learning_rate=0.001, max_bin=256, max_cat_to_onehot=4,
max_delta_step=0, max_depth=8, max_leaves=0, min_child_weight=1,
missing=nan, monotone_constraints='()', n_estimators=100,
n_jobs=0, num_parallel_tree=1, predictor='auto', random_state=42,
reg_alpha=0, reg_lambda=1, ...)In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. XGBClassifier(base_score=0.5, booster='gbtree', callbacks=None,
colsample_bylevel=1, colsample_bynode=1, colsample_bytree=1,
early_stopping_rounds=None, enable_categorical=False,
eval_metric=None, gamma=0, gpu_id=-1, grow_policy='depthwise',
importance_type=None, interaction_constraints='',
learning_rate=0.001, max_bin=256, max_cat_to_onehot=4,
max_delta_step=0, max_depth=8, max_leaves=0, min_child_weight=1,
missing=nan, monotone_constraints='()', n_estimators=100,
n_jobs=0, num_parallel_tree=1, predictor='auto', random_state=42,
reg_alpha=0, reg_lambda=1, ...)#xgb_classifier5(xgb_grid.best_estimator_
# training set
y_pred_xgb_tr555 = xgb_random_search5.predict(X_train55)
accuracy_xgb_tr555 = accuracy_score(y_train55, y_pred_xgb_tr555)
print("XGBoost Model Score (Training Set): %.2f%%" % (accuracy_xgb_tr555 * 100.0))
# testing set
y_pred_xgb_t555 = xgb_random_search5.predict(X_test55)
xgb_predictions555 = [round(value) for value in y_pred_xgb_t555]
# evaluate predictions
accuracy_xgb_t555 = accuracy_score(y_test55, xgb_predictions555)
print("XGBoost Model Score (Testing Set): %.2f%%" % (accuracy_xgb_t555 * 100.0))
XGBoost Model Score (Training Set): 80.40% XGBoost Model Score (Testing Set): 73.84%
plot_confusion_matrix(xgb_random_search5, X_test55, y_test55,cmap="YlGnBu"); # good improvement
# Printing the classification report
print(classification_report(y_test55, y_pred_xgb_t555))
precision recall f1-score support
0 0.82 0.76 0.79 318
1 0.62 0.70 0.66 179
accuracy 0.74 497
macro avg 0.72 0.73 0.72 497
weighted avg 0.75 0.74 0.74 497
# Accuracy Score for each company.
xgb_accuracy11 = {'Aramco': accuracy_xgb_t111*100, 'SARCO': accuracy_xgb_t222*100, 'Rabigh': accuracy_xgb_t333*100, 'Bahri':accuracy_xgb_t444*100, 'Aldrees': accuracy_xgb_t555*100}
print(xgb_accuracy11)
xgb_names11 = list(xgb_accuracy11.keys())
xgb_values11 = list(xgb_accuracy11.values())
fig, ax = plt.subplots(figsize=(15,8))
plt.title('Optimized XGBoost Accuracy Score per Company (RandomizedSearch CV)')
bars = ax.barh(xgb_names11, xgb_values11)
ax.bar_label(bars, fmt='%.2f%%')
plt.show()
{'Aramco': 67.91044776119402, 'SARCO': 94.56740442655935, 'Rabigh': 96.579476861167, 'Bahri': 67.20321931589537, 'Aldrees': 73.8430583501006}
xgboost = xgb.XGBClassifier(objective= "binary:logistic", random_state=42)
#brute force scan for all parameters, here are the tricks
#usually max_depth is 6,7,8
#learning rate is around 0.05, but small changes may make big diff
#tuning min_child_weight subsample colsample_bytree can have
#much fun of fighting against overfit
#n_estimators is how many round of boosting
#finally, ensemble xgboost with multiple seeds may reduce variance
parameters = {'base_score':[0.5, 0.7],
'learning_rate': [0.01, 0.02, 0.04, 0.05, 0.1],
'max_depth': [2, 4, 5, 6, 8],
'subsample': [0.2, 0.3, 0.5, 0.7, 0.8, 1],
'colsample_bytree': [0.3, 0.7, 1, 2, 3],
'n_estimators': [100],
'scale_pos_weight': [estimate1]#, estimate2, estimate3, estimate4, estimate5]
} #number of trees
xgb_grid = GridSearchCV(xgboost, parameters, n_jobs=-1,
cv=KFold(n_splits=5, shuffle=True, random_state=42), # at least 5 splits
scoring='accuracy',
verbose=2)
xgb_grid
GridSearchCV(cv=KFold(n_splits=5, random_state=42, shuffle=True),
estimator=XGBClassifier(base_score=None, booster=None,
callbacks=None, colsample_bylevel=None,
colsample_bynode=None,
colsample_bytree=None,
early_stopping_rounds=None,
enable_categorical=False, eval_metric=None,
gamma=None, gpu_id=None, grow_policy=None,
importance_type=None,
interaction_constraint...
num_parallel_tree=None, predictor=None,
random_state=42, reg_alpha=None,
reg_lambda=None, ...),
n_jobs=-1,
param_grid={'base_score': [0.5, 0.7],
'colsample_bytree': [0.3, 0.7, 1, 2, 3],
'learning_rate': [0.01, 0.02, 0.04, 0.05, 0.1],
'max_depth': [2, 4, 5, 6, 8], 'n_estimators': [100],
'scale_pos_weight': [0.7631578947368421],
'subsample': [0.2, 0.3, 0.5, 0.7, 0.8, 1]},
scoring='accuracy', verbose=2)In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. GridSearchCV(cv=KFold(n_splits=5, random_state=42, shuffle=True),
estimator=XGBClassifier(base_score=None, booster=None,
callbacks=None, colsample_bylevel=None,
colsample_bynode=None,
colsample_bytree=None,
early_stopping_rounds=None,
enable_categorical=False, eval_metric=None,
gamma=None, gpu_id=None, grow_policy=None,
importance_type=None,
interaction_constraint...
num_parallel_tree=None, predictor=None,
random_state=42, reg_alpha=None,
reg_lambda=None, ...),
n_jobs=-1,
param_grid={'base_score': [0.5, 0.7],
'colsample_bytree': [0.3, 0.7, 1, 2, 3],
'learning_rate': [0.01, 0.02, 0.04, 0.05, 0.1],
'max_depth': [2, 4, 5, 6, 8], 'n_estimators': [100],
'scale_pos_weight': [0.7631578947368421],
'subsample': [0.2, 0.3, 0.5, 0.7, 0.8, 1]},
scoring='accuracy', verbose=2)XGBClassifier(base_score=None, booster=None, callbacks=None,
colsample_bylevel=None, colsample_bynode=None,
colsample_bytree=None, early_stopping_rounds=None,
enable_categorical=False, eval_metric=None, gamma=None,
gpu_id=None, grow_policy=None, importance_type=None,
interaction_constraints=None, learning_rate=None, max_bin=None,
max_cat_to_onehot=None, max_delta_step=None, max_depth=None,
max_leaves=None, min_child_weight=None, missing=nan,
monotone_constraints=None, n_estimators=100, n_jobs=None,
num_parallel_tree=None, predictor=None, random_state=42,
reg_alpha=None, reg_lambda=None, ...)XGBClassifier(base_score=None, booster=None, callbacks=None,
colsample_bylevel=None, colsample_bynode=None,
colsample_bytree=None, early_stopping_rounds=None,
enable_categorical=False, eval_metric=None, gamma=None,
gpu_id=None, grow_policy=None, importance_type=None,
interaction_constraints=None, learning_rate=None, max_bin=None,
max_cat_to_onehot=None, max_delta_step=None, max_depth=None,
max_leaves=None, min_child_weight=None, missing=nan,
monotone_constraints=None, n_estimators=100, n_jobs=None,
num_parallel_tree=None, predictor=None, random_state=42,
reg_alpha=None, reg_lambda=None, ...)X_train11, X_test11 = scale_data(X_train11, X_test11)
xgb_grid_result1 = xgb_grid.fit(X_train11, y_train11)
xgb_grid_result1
Fitting 5 folds for each of 1500 candidates, totalling 7500 fits
GridSearchCV(cv=KFold(n_splits=5, random_state=42, shuffle=True),
estimator=XGBClassifier(base_score=None, booster=None,
callbacks=None, colsample_bylevel=None,
colsample_bynode=None,
colsample_bytree=None,
early_stopping_rounds=None,
enable_categorical=False, eval_metric=None,
gamma=None, gpu_id=None, grow_policy=None,
importance_type=None,
interaction_constraint...
num_parallel_tree=None, predictor=None,
random_state=42, reg_alpha=None,
reg_lambda=None, ...),
n_jobs=-1,
param_grid={'base_score': [0.5, 0.7],
'colsample_bytree': [0.3, 0.7, 1, 2, 3],
'learning_rate': [0.01, 0.02, 0.04, 0.05, 0.1],
'max_depth': [2, 4, 5, 6, 8], 'n_estimators': [100],
'scale_pos_weight': [0.7631578947368421],
'subsample': [0.2, 0.3, 0.5, 0.7, 0.8, 1]},
scoring='accuracy', verbose=2)In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. GridSearchCV(cv=KFold(n_splits=5, random_state=42, shuffle=True),
estimator=XGBClassifier(base_score=None, booster=None,
callbacks=None, colsample_bylevel=None,
colsample_bynode=None,
colsample_bytree=None,
early_stopping_rounds=None,
enable_categorical=False, eval_metric=None,
gamma=None, gpu_id=None, grow_policy=None,
importance_type=None,
interaction_constraint...
num_parallel_tree=None, predictor=None,
random_state=42, reg_alpha=None,
reg_lambda=None, ...),
n_jobs=-1,
param_grid={'base_score': [0.5, 0.7],
'colsample_bytree': [0.3, 0.7, 1, 2, 3],
'learning_rate': [0.01, 0.02, 0.04, 0.05, 0.1],
'max_depth': [2, 4, 5, 6, 8], 'n_estimators': [100],
'scale_pos_weight': [0.7631578947368421],
'subsample': [0.2, 0.3, 0.5, 0.7, 0.8, 1]},
scoring='accuracy', verbose=2)XGBClassifier(base_score=None, booster=None, callbacks=None,
colsample_bylevel=None, colsample_bynode=None,
colsample_bytree=None, early_stopping_rounds=None,
enable_categorical=False, eval_metric=None, gamma=None,
gpu_id=None, grow_policy=None, importance_type=None,
interaction_constraints=None, learning_rate=None, max_bin=None,
max_cat_to_onehot=None, max_delta_step=None, max_depth=None,
max_leaves=None, min_child_weight=None, missing=nan,
monotone_constraints=None, n_estimators=100, n_jobs=None,
num_parallel_tree=None, predictor=None, random_state=42,
reg_alpha=None, reg_lambda=None, ...)XGBClassifier(base_score=None, booster=None, callbacks=None,
colsample_bylevel=None, colsample_bynode=None,
colsample_bytree=None, early_stopping_rounds=None,
enable_categorical=False, eval_metric=None, gamma=None,
gpu_id=None, grow_policy=None, importance_type=None,
interaction_constraints=None, learning_rate=None, max_bin=None,
max_cat_to_onehot=None, max_delta_step=None, max_depth=None,
max_leaves=None, min_child_weight=None, missing=nan,
monotone_constraints=None, n_estimators=100, n_jobs=None,
num_parallel_tree=None, predictor=None, random_state=42,
reg_alpha=None, reg_lambda=None, ...)print("\n\nBest score: %f using %s" % (xgb_grid_result1.best_score_, xgb_grid_result1.best_params_))
print("Best Estimator: ")
xgb_grid.best_estimator_
Best score: 0.643683 using {'base_score': 0.5, 'colsample_bytree': 1, 'learning_rate': 0.05, 'max_depth': 2, 'n_estimators': 100, 'scale_pos_weight': 0.7631578947368421, 'subsample': 0.2}
Best Estimator:
XGBClassifier(base_score=0.5, booster='gbtree', callbacks=None,
colsample_bylevel=1, colsample_bynode=1, colsample_bytree=1,
early_stopping_rounds=None, enable_categorical=False,
eval_metric=None, gamma=0, gpu_id=-1, grow_policy='depthwise',
importance_type=None, interaction_constraints='',
learning_rate=0.05, max_bin=256, max_cat_to_onehot=4,
max_delta_step=0, max_depth=2, max_leaves=0, min_child_weight=1,
missing=nan, monotone_constraints='()', n_estimators=100,
n_jobs=0, num_parallel_tree=1, predictor='auto', random_state=42,
reg_alpha=0, reg_lambda=1, ...)In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. XGBClassifier(base_score=0.5, booster='gbtree', callbacks=None,
colsample_bylevel=1, colsample_bynode=1, colsample_bytree=1,
early_stopping_rounds=None, enable_categorical=False,
eval_metric=None, gamma=0, gpu_id=-1, grow_policy='depthwise',
importance_type=None, interaction_constraints='',
learning_rate=0.05, max_bin=256, max_cat_to_onehot=4,
max_delta_step=0, max_depth=2, max_leaves=0, min_child_weight=1,
missing=nan, monotone_constraints='()', n_estimators=100,
n_jobs=0, num_parallel_tree=1, predictor='auto', random_state=42,
reg_alpha=0, reg_lambda=1, ...)# training set
y_pred_xgb_tr11 = xgb_grid.predict(X_train11)
accuracy_xgb_tr11 = accuracy_score(y_train11, y_pred_xgb_tr11)
print("XGBoost Model Score (Training Set): %.2f%%" % (accuracy_xgb_tr11 * 100.0))
# testing set
y_pred_xgb_t11 = xgb_grid.predict(X_test11)
xgb_predictions11 = [round(value) for value in y_pred_xgb_t11]
# evaluate predictions
accuracy_xgb_t11 = accuracy_score(y_test11, xgb_predictions11)
print("XGBoost Model Score (Testing Set): %.2f%%" % (accuracy_xgb_t11 * 100.0))
XGBoost Model Score (Training Set): 72.39% XGBoost Model Score (Testing Set): 62.69%
plot_confusion_matrix(xgb_grid, X_test11, y_test11,cmap="YlGnBu"); # good improvement
# Printing the classification report
print(classification_report(y_test11, y_pred_xgb_t11))
precision recall f1-score support
0 0.53 0.57 0.55 54
1 0.70 0.66 0.68 80
accuracy 0.63 134
macro avg 0.62 0.62 0.62 134
weighted avg 0.63 0.63 0.63 134
parameters['scale_pos_weight'] = [estimate2]
xgb_grid2 = GridSearchCV(xgboost, parameters, n_jobs=-1,
cv=KFold(n_splits=5, shuffle=True, random_state=42), # at least 5 splits
scoring='accuracy',
verbose=2)
xgb_grid_result2 = xgb_grid2.fit(X_train22, y_train22)
xgb_grid_result2
Fitting 5 folds for each of 1500 candidates, totalling 7500 fits
GridSearchCV(cv=KFold(n_splits=5, random_state=42, shuffle=True),
estimator=XGBClassifier(base_score=None, booster=None,
callbacks=None, colsample_bylevel=None,
colsample_bynode=None,
colsample_bytree=None,
early_stopping_rounds=None,
enable_categorical=False, eval_metric=None,
gamma=None, gpu_id=None, grow_policy=None,
importance_type=None,
interaction_constraint...
num_parallel_tree=None, predictor=None,
random_state=42, reg_alpha=None,
reg_lambda=None, ...),
n_jobs=-1,
param_grid={'base_score': [0.5, 0.7],
'colsample_bytree': [0.3, 0.7, 1, 2, 3],
'learning_rate': [0.01, 0.02, 0.04, 0.05, 0.1],
'max_depth': [2, 4, 5, 6, 8], 'n_estimators': [100],
'scale_pos_weight': [16.110344827586207],
'subsample': [0.2, 0.3, 0.5, 0.7, 0.8, 1]},
scoring='accuracy', verbose=2)In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. GridSearchCV(cv=KFold(n_splits=5, random_state=42, shuffle=True),
estimator=XGBClassifier(base_score=None, booster=None,
callbacks=None, colsample_bylevel=None,
colsample_bynode=None,
colsample_bytree=None,
early_stopping_rounds=None,
enable_categorical=False, eval_metric=None,
gamma=None, gpu_id=None, grow_policy=None,
importance_type=None,
interaction_constraint...
num_parallel_tree=None, predictor=None,
random_state=42, reg_alpha=None,
reg_lambda=None, ...),
n_jobs=-1,
param_grid={'base_score': [0.5, 0.7],
'colsample_bytree': [0.3, 0.7, 1, 2, 3],
'learning_rate': [0.01, 0.02, 0.04, 0.05, 0.1],
'max_depth': [2, 4, 5, 6, 8], 'n_estimators': [100],
'scale_pos_weight': [16.110344827586207],
'subsample': [0.2, 0.3, 0.5, 0.7, 0.8, 1]},
scoring='accuracy', verbose=2)XGBClassifier(base_score=None, booster=None, callbacks=None,
colsample_bylevel=None, colsample_bynode=None,
colsample_bytree=None, early_stopping_rounds=None,
enable_categorical=False, eval_metric=None, gamma=None,
gpu_id=None, grow_policy=None, importance_type=None,
interaction_constraints=None, learning_rate=None, max_bin=None,
max_cat_to_onehot=None, max_delta_step=None, max_depth=None,
max_leaves=None, min_child_weight=None, missing=nan,
monotone_constraints=None, n_estimators=100, n_jobs=None,
num_parallel_tree=None, predictor=None, random_state=42,
reg_alpha=None, reg_lambda=None, ...)XGBClassifier(base_score=None, booster=None, callbacks=None,
colsample_bylevel=None, colsample_bynode=None,
colsample_bytree=None, early_stopping_rounds=None,
enable_categorical=False, eval_metric=None, gamma=None,
gpu_id=None, grow_policy=None, importance_type=None,
interaction_constraints=None, learning_rate=None, max_bin=None,
max_cat_to_onehot=None, max_delta_step=None, max_depth=None,
max_leaves=None, min_child_weight=None, missing=nan,
monotone_constraints=None, n_estimators=100, n_jobs=None,
num_parallel_tree=None, predictor=None, random_state=42,
reg_alpha=None, reg_lambda=None, ...)print("\n\nBest score: %f using %s" % (xgb_grid_result2.best_score_, xgb_grid_result2.best_params_))
print("Best Estimator: ")
xgb_grid2.best_estimator_
Best score: 0.936487 using {'base_score': 0.7, 'colsample_bytree': 0.3, 'learning_rate': 0.1, 'max_depth': 8, 'n_estimators': 100, 'scale_pos_weight': 16.110344827586207, 'subsample': 1}
Best Estimator:
XGBClassifier(base_score=0.7, booster='gbtree', callbacks=None,
colsample_bylevel=1, colsample_bynode=1, colsample_bytree=0.3,
early_stopping_rounds=None, enable_categorical=False,
eval_metric=None, gamma=0, gpu_id=-1, grow_policy='depthwise',
importance_type=None, interaction_constraints='',
learning_rate=0.1, max_bin=256, max_cat_to_onehot=4,
max_delta_step=0, max_depth=8, max_leaves=0, min_child_weight=1,
missing=nan, monotone_constraints='()', n_estimators=100,
n_jobs=0, num_parallel_tree=1, predictor='auto', random_state=42,
reg_alpha=0, reg_lambda=1, ...)In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. XGBClassifier(base_score=0.7, booster='gbtree', callbacks=None,
colsample_bylevel=1, colsample_bynode=1, colsample_bytree=0.3,
early_stopping_rounds=None, enable_categorical=False,
eval_metric=None, gamma=0, gpu_id=-1, grow_policy='depthwise',
importance_type=None, interaction_constraints='',
learning_rate=0.1, max_bin=256, max_cat_to_onehot=4,
max_delta_step=0, max_depth=8, max_leaves=0, min_child_weight=1,
missing=nan, monotone_constraints='()', n_estimators=100,
n_jobs=0, num_parallel_tree=1, predictor='auto', random_state=42,
reg_alpha=0, reg_lambda=1, ...)# training set
y_pred_xgb_tr22 = xgb_grid2.predict(X_train22)
accuracy_xgb_tr22 = accuracy_score(y_train22, y_pred_xgb_tr22)
print("XGBoost Model Score (Training Set): %.2f%%" % (accuracy_xgb_tr22 * 100.0))
# testing set
y_pred_xgb_t22 = xgb_grid2.predict(X_test22)
xgb_predictions22 = [round(value) for value in y_pred_xgb_t22]
# evaluate predictions
accuracy_xgb_t22 = accuracy_score(y_test2, xgb_predictions22)
print("XGBoost Model Score (Testing Set): %.2f%%" % (accuracy_xgb_t22 * 100.0))
XGBoost Model Score (Training Set): 100.00% XGBoost Model Score (Testing Set): 94.16%
plot_confusion_matrix(xgb_grid2, X_test22, y_test22,cmap="YlGnBu"); # good improvement
# Printing the classification report
print(classification_report(y_test22, y_pred_xgb_t22))
precision recall f1-score support
0 0.95 0.99 0.97 468
1 0.50 0.24 0.33 29
accuracy 0.94 497
macro avg 0.73 0.61 0.65 497
weighted avg 0.93 0.94 0.93 497
parameters['scale_pos_weight'] = [estimate3]
xgb_grid3 = GridSearchCV(xgboost, parameters, n_jobs=-1,
cv=KFold(n_splits=5, shuffle=True, random_state=42), # at least 5 splits
scoring='accuracy',
verbose=2)
xgb_grid_result3 = xgb_grid3.fit(X_train33, y_train33)
xgb_grid_result3
Fitting 5 folds for each of 1500 candidates, totalling 7500 fits
GridSearchCV(cv=KFold(n_splits=5, random_state=42, shuffle=True),
estimator=XGBClassifier(base_score=None, booster=None,
callbacks=None, colsample_bylevel=None,
colsample_bynode=None,
colsample_bytree=None,
early_stopping_rounds=None,
enable_categorical=False, eval_metric=None,
gamma=None, gpu_id=None, grow_policy=None,
importance_type=None,
interaction_constraint...
num_parallel_tree=None, predictor=None,
random_state=42, reg_alpha=None,
reg_lambda=None, ...),
n_jobs=-1,
param_grid={'base_score': [0.5, 0.7],
'colsample_bytree': [0.3, 0.7, 1, 2, 3],
'learning_rate': [0.01, 0.02, 0.04, 0.05, 0.1],
'max_depth': [2, 4, 5, 6, 8], 'n_estimators': [100],
'scale_pos_weight': [0.0324459234608985],
'subsample': [0.2, 0.3, 0.5, 0.7, 0.8, 1]},
scoring='accuracy', verbose=2)In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. GridSearchCV(cv=KFold(n_splits=5, random_state=42, shuffle=True),
estimator=XGBClassifier(base_score=None, booster=None,
callbacks=None, colsample_bylevel=None,
colsample_bynode=None,
colsample_bytree=None,
early_stopping_rounds=None,
enable_categorical=False, eval_metric=None,
gamma=None, gpu_id=None, grow_policy=None,
importance_type=None,
interaction_constraint...
num_parallel_tree=None, predictor=None,
random_state=42, reg_alpha=None,
reg_lambda=None, ...),
n_jobs=-1,
param_grid={'base_score': [0.5, 0.7],
'colsample_bytree': [0.3, 0.7, 1, 2, 3],
'learning_rate': [0.01, 0.02, 0.04, 0.05, 0.1],
'max_depth': [2, 4, 5, 6, 8], 'n_estimators': [100],
'scale_pos_weight': [0.0324459234608985],
'subsample': [0.2, 0.3, 0.5, 0.7, 0.8, 1]},
scoring='accuracy', verbose=2)XGBClassifier(base_score=None, booster=None, callbacks=None,
colsample_bylevel=None, colsample_bynode=None,
colsample_bytree=None, early_stopping_rounds=None,
enable_categorical=False, eval_metric=None, gamma=None,
gpu_id=None, grow_policy=None, importance_type=None,
interaction_constraints=None, learning_rate=None, max_bin=None,
max_cat_to_onehot=None, max_delta_step=None, max_depth=None,
max_leaves=None, min_child_weight=None, missing=nan,
monotone_constraints=None, n_estimators=100, n_jobs=None,
num_parallel_tree=None, predictor=None, random_state=42,
reg_alpha=None, reg_lambda=None, ...)XGBClassifier(base_score=None, booster=None, callbacks=None,
colsample_bylevel=None, colsample_bynode=None,
colsample_bytree=None, early_stopping_rounds=None,
enable_categorical=False, eval_metric=None, gamma=None,
gpu_id=None, grow_policy=None, importance_type=None,
interaction_constraints=None, learning_rate=None, max_bin=None,
max_cat_to_onehot=None, max_delta_step=None, max_depth=None,
max_leaves=None, min_child_weight=None, missing=nan,
monotone_constraints=None, n_estimators=100, n_jobs=None,
num_parallel_tree=None, predictor=None, random_state=42,
reg_alpha=None, reg_lambda=None, ...)print("\n\nBest score: %f using %s" % (xgb_grid_result3.best_score_, xgb_grid_result3.best_params_))
print("Best Estimator: ")
xgb_grid3.best_estimator_
Best score: 0.969270 using {'base_score': 0.7, 'colsample_bytree': 0.3, 'learning_rate': 0.01, 'max_depth': 2, 'n_estimators': 100, 'scale_pos_weight': 0.0324459234608985, 'subsample': 0.2}
Best Estimator:
XGBClassifier(base_score=0.7, booster='gbtree', callbacks=None,
colsample_bylevel=1, colsample_bynode=1, colsample_bytree=0.3,
early_stopping_rounds=None, enable_categorical=False,
eval_metric=None, gamma=0, gpu_id=-1, grow_policy='depthwise',
importance_type=None, interaction_constraints='',
learning_rate=0.01, max_bin=256, max_cat_to_onehot=4,
max_delta_step=0, max_depth=2, max_leaves=0, min_child_weight=1,
missing=nan, monotone_constraints='()', n_estimators=100,
n_jobs=0, num_parallel_tree=1, predictor='auto', random_state=42,
reg_alpha=0, reg_lambda=1, ...)In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. XGBClassifier(base_score=0.7, booster='gbtree', callbacks=None,
colsample_bylevel=1, colsample_bynode=1, colsample_bytree=0.3,
early_stopping_rounds=None, enable_categorical=False,
eval_metric=None, gamma=0, gpu_id=-1, grow_policy='depthwise',
importance_type=None, interaction_constraints='',
learning_rate=0.01, max_bin=256, max_cat_to_onehot=4,
max_delta_step=0, max_depth=2, max_leaves=0, min_child_weight=1,
missing=nan, monotone_constraints='()', n_estimators=100,
n_jobs=0, num_parallel_tree=1, predictor='auto', random_state=42,
reg_alpha=0, reg_lambda=1, ...)# training set
y_pred_xgb_tr33 = xgb_grid3.predict(X_train33)
accuracy_xgb_tr33 = accuracy_score(y_train33, y_pred_xgb_tr33)
print("XGBoost Model Score (Training Set): %.2f%%" % (accuracy_xgb_tr33 * 100.0))
# testing set
y_pred_xgb_t33 = xgb_grid3.predict(X_test33)
xgb_predictions33 = [round(value) for value in y_pred_xgb_t33]
# evaluate predictions
accuracy_xgb_t33 = accuracy_score(y_test33, xgb_predictions33)
print("XGBoost Model Score (Testing Set): %.2f%%" % (accuracy_xgb_t33 * 100.0))
XGBoost Model Score (Training Set): 96.93% XGBoost Model Score (Testing Set): 96.58%
plot_confusion_matrix(xgb_grid3, X_test33, y_test33,cmap="YlGnBu"); # good improvement
# Printing the classification report
print(classification_report(y_test33, y_pred_xgb_t33))
precision recall f1-score support
0 0.00 0.00 0.00 17
1 0.97 1.00 0.98 480
accuracy 0.97 497
macro avg 0.48 0.50 0.49 497
weighted avg 0.93 0.97 0.95 497
parameters['scale_pos_weight'] = [estimate4]
xgb_grid4 = GridSearchCV(xgboost, parameters, n_jobs=-1,
cv=KFold(n_splits=5, shuffle=True, random_state=42), # at least 5 splits
scoring='accuracy',
verbose=2)
xgb_grid_result4 = xgb_grid4.fit(X_train44, y_train44)
xgb_grid_result4
Fitting 5 folds for each of 1500 candidates, totalling 7500 fits
GridSearchCV(cv=KFold(n_splits=5, random_state=42, shuffle=True),
estimator=XGBClassifier(base_score=None, booster=None,
callbacks=None, colsample_bylevel=None,
colsample_bynode=None,
colsample_bytree=None,
early_stopping_rounds=None,
enable_categorical=False, eval_metric=None,
gamma=None, gpu_id=None, grow_policy=None,
importance_type=None,
interaction_constraint...
num_parallel_tree=None, predictor=None,
random_state=42, reg_alpha=None,
reg_lambda=None, ...),
n_jobs=-1,
param_grid={'base_score': [0.5, 0.7],
'colsample_bytree': [0.3, 0.7, 1, 2, 3],
'learning_rate': [0.01, 0.02, 0.04, 0.05, 0.1],
'max_depth': [2, 4, 5, 6, 8], 'n_estimators': [100],
'scale_pos_weight': [0.7508821453775583],
'subsample': [0.2, 0.3, 0.5, 0.7, 0.8, 1]},
scoring='accuracy', verbose=2)In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. GridSearchCV(cv=KFold(n_splits=5, random_state=42, shuffle=True),
estimator=XGBClassifier(base_score=None, booster=None,
callbacks=None, colsample_bylevel=None,
colsample_bynode=None,
colsample_bytree=None,
early_stopping_rounds=None,
enable_categorical=False, eval_metric=None,
gamma=None, gpu_id=None, grow_policy=None,
importance_type=None,
interaction_constraint...
num_parallel_tree=None, predictor=None,
random_state=42, reg_alpha=None,
reg_lambda=None, ...),
n_jobs=-1,
param_grid={'base_score': [0.5, 0.7],
'colsample_bytree': [0.3, 0.7, 1, 2, 3],
'learning_rate': [0.01, 0.02, 0.04, 0.05, 0.1],
'max_depth': [2, 4, 5, 6, 8], 'n_estimators': [100],
'scale_pos_weight': [0.7508821453775583],
'subsample': [0.2, 0.3, 0.5, 0.7, 0.8, 1]},
scoring='accuracy', verbose=2)XGBClassifier(base_score=None, booster=None, callbacks=None,
colsample_bylevel=None, colsample_bynode=None,
colsample_bytree=None, early_stopping_rounds=None,
enable_categorical=False, eval_metric=None, gamma=None,
gpu_id=None, grow_policy=None, importance_type=None,
interaction_constraints=None, learning_rate=None, max_bin=None,
max_cat_to_onehot=None, max_delta_step=None, max_depth=None,
max_leaves=None, min_child_weight=None, missing=nan,
monotone_constraints=None, n_estimators=100, n_jobs=None,
num_parallel_tree=None, predictor=None, random_state=42,
reg_alpha=None, reg_lambda=None, ...)XGBClassifier(base_score=None, booster=None, callbacks=None,
colsample_bylevel=None, colsample_bynode=None,
colsample_bytree=None, early_stopping_rounds=None,
enable_categorical=False, eval_metric=None, gamma=None,
gpu_id=None, grow_policy=None, importance_type=None,
interaction_constraints=None, learning_rate=None, max_bin=None,
max_cat_to_onehot=None, max_delta_step=None, max_depth=None,
max_leaves=None, min_child_weight=None, missing=nan,
monotone_constraints=None, n_estimators=100, n_jobs=None,
num_parallel_tree=None, predictor=None, random_state=42,
reg_alpha=None, reg_lambda=None, ...)print("\n\nBest score: %f using %s" % (xgb_grid_result4.best_score_, xgb_grid_result4.best_params_))
print("Best Estimator: ")
xgb_grid4.best_estimator_
Best score: 0.705127 using {'base_score': 0.7, 'colsample_bytree': 0.7, 'learning_rate': 0.02, 'max_depth': 5, 'n_estimators': 100, 'scale_pos_weight': 0.7508821453775583, 'subsample': 0.3}
Best Estimator:
XGBClassifier(base_score=0.7, booster='gbtree', callbacks=None,
colsample_bylevel=1, colsample_bynode=1, colsample_bytree=0.7,
early_stopping_rounds=None, enable_categorical=False,
eval_metric=None, gamma=0, gpu_id=-1, grow_policy='depthwise',
importance_type=None, interaction_constraints='',
learning_rate=0.02, max_bin=256, max_cat_to_onehot=4,
max_delta_step=0, max_depth=5, max_leaves=0, min_child_weight=1,
missing=nan, monotone_constraints='()', n_estimators=100,
n_jobs=0, num_parallel_tree=1, predictor='auto', random_state=42,
reg_alpha=0, reg_lambda=1, ...)In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. XGBClassifier(base_score=0.7, booster='gbtree', callbacks=None,
colsample_bylevel=1, colsample_bynode=1, colsample_bytree=0.7,
early_stopping_rounds=None, enable_categorical=False,
eval_metric=None, gamma=0, gpu_id=-1, grow_policy='depthwise',
importance_type=None, interaction_constraints='',
learning_rate=0.02, max_bin=256, max_cat_to_onehot=4,
max_delta_step=0, max_depth=5, max_leaves=0, min_child_weight=1,
missing=nan, monotone_constraints='()', n_estimators=100,
n_jobs=0, num_parallel_tree=1, predictor='auto', random_state=42,
reg_alpha=0, reg_lambda=1, ...)# training set
y_pred_xgb_tr44 = xgb_grid4.predict(X_train44)
accuracy_xgb_tr44 = accuracy_score(y_train44, y_pred_xgb_tr44)
print("XGBoost Model Score (Training Set): %.2f%%" % (accuracy_xgb_tr44 * 100.0))
# testing set
y_pred_xgb_t44 = xgb_grid4.predict(X_test44)
xgb_predictions44 = [round(value) for value in y_pred_xgb_t44]
# evaluate predictions
accuracy_xgb_t44 = accuracy_score(y_test44, xgb_predictions44)
print("XGBoost Model Score (Testing Set): %.2f%%" % (accuracy_xgb_t44 * 100.0))
XGBoost Model Score (Training Set): 75.71% XGBoost Model Score (Testing Set): 66.40%
plot_confusion_matrix(xgb_grid4, X_test44, y_test44,cmap="YlGnBu"); # good improvement
# Printing the classification report
print(classification_report(y_test44, y_pred_xgb_t44))
precision recall f1-score support
0 0.64 0.55 0.60 222
1 0.68 0.75 0.71 275
accuracy 0.66 497
macro avg 0.66 0.65 0.65 497
weighted avg 0.66 0.66 0.66 497
parameters['scale_pos_weight'] = [estimate5]
xgb_grid5 = GridSearchCV(xgboost, parameters, n_jobs=-1,
cv=KFold(n_splits=5, shuffle=True, random_state=42), # at least 5 splits
scoring='accuracy',
verbose=2)
xgb_grid_result5 = xgb_grid5.fit(X_train55, y_train55)
xgb_grid_result5
Fitting 5 folds for each of 1500 candidates, totalling 7500 fits
GridSearchCV(cv=KFold(n_splits=5, random_state=42, shuffle=True),
estimator=XGBClassifier(base_score=None, booster=None,
callbacks=None, colsample_bylevel=None,
colsample_bynode=None,
colsample_bytree=None,
early_stopping_rounds=None,
enable_categorical=False, eval_metric=None,
gamma=None, gpu_id=None, grow_policy=None,
importance_type=None,
interaction_constraint...
num_parallel_tree=None, predictor=None,
random_state=42, reg_alpha=None,
reg_lambda=None, ...),
n_jobs=-1,
param_grid={'base_score': [0.5, 0.7],
'colsample_bytree': [0.3, 0.7, 1, 2, 3],
'learning_rate': [0.01, 0.02, 0.04, 0.05, 0.1],
'max_depth': [2, 4, 5, 6, 8], 'n_estimators': [100],
'scale_pos_weight': [1.7486157253599115],
'subsample': [0.2, 0.3, 0.5, 0.7, 0.8, 1]},
scoring='accuracy', verbose=2)In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. GridSearchCV(cv=KFold(n_splits=5, random_state=42, shuffle=True),
estimator=XGBClassifier(base_score=None, booster=None,
callbacks=None, colsample_bylevel=None,
colsample_bynode=None,
colsample_bytree=None,
early_stopping_rounds=None,
enable_categorical=False, eval_metric=None,
gamma=None, gpu_id=None, grow_policy=None,
importance_type=None,
interaction_constraint...
num_parallel_tree=None, predictor=None,
random_state=42, reg_alpha=None,
reg_lambda=None, ...),
n_jobs=-1,
param_grid={'base_score': [0.5, 0.7],
'colsample_bytree': [0.3, 0.7, 1, 2, 3],
'learning_rate': [0.01, 0.02, 0.04, 0.05, 0.1],
'max_depth': [2, 4, 5, 6, 8], 'n_estimators': [100],
'scale_pos_weight': [1.7486157253599115],
'subsample': [0.2, 0.3, 0.5, 0.7, 0.8, 1]},
scoring='accuracy', verbose=2)XGBClassifier(base_score=None, booster=None, callbacks=None,
colsample_bylevel=None, colsample_bynode=None,
colsample_bytree=None, early_stopping_rounds=None,
enable_categorical=False, eval_metric=None, gamma=None,
gpu_id=None, grow_policy=None, importance_type=None,
interaction_constraints=None, learning_rate=None, max_bin=None,
max_cat_to_onehot=None, max_delta_step=None, max_depth=None,
max_leaves=None, min_child_weight=None, missing=nan,
monotone_constraints=None, n_estimators=100, n_jobs=None,
num_parallel_tree=None, predictor=None, random_state=42,
reg_alpha=None, reg_lambda=None, ...)XGBClassifier(base_score=None, booster=None, callbacks=None,
colsample_bylevel=None, colsample_bynode=None,
colsample_bytree=None, early_stopping_rounds=None,
enable_categorical=False, eval_metric=None, gamma=None,
gpu_id=None, grow_policy=None, importance_type=None,
interaction_constraints=None, learning_rate=None, max_bin=None,
max_cat_to_onehot=None, max_delta_step=None, max_depth=None,
max_leaves=None, min_child_weight=None, missing=nan,
monotone_constraints=None, n_estimators=100, n_jobs=None,
num_parallel_tree=None, predictor=None, random_state=42,
reg_alpha=None, reg_lambda=None, ...)print("\n\nBest score: %f using %s" % (xgb_grid_result5.best_score_, xgb_grid_result5.best_params_))
print("Best Estimator: ")
xgb_grid5.best_estimator_
Best score: 0.714358 using {'base_score': 0.7, 'colsample_bytree': 0.7, 'learning_rate': 0.05, 'max_depth': 6, 'n_estimators': 100, 'scale_pos_weight': 1.7486157253599115, 'subsample': 0.7}
Best Estimator:
XGBClassifier(base_score=0.7, booster='gbtree', callbacks=None,
colsample_bylevel=1, colsample_bynode=1, colsample_bytree=0.7,
early_stopping_rounds=None, enable_categorical=False,
eval_metric=None, gamma=0, gpu_id=-1, grow_policy='depthwise',
importance_type=None, interaction_constraints='',
learning_rate=0.05, max_bin=256, max_cat_to_onehot=4,
max_delta_step=0, max_depth=6, max_leaves=0, min_child_weight=1,
missing=nan, monotone_constraints='()', n_estimators=100,
n_jobs=0, num_parallel_tree=1, predictor='auto', random_state=42,
reg_alpha=0, reg_lambda=1, ...)In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. XGBClassifier(base_score=0.7, booster='gbtree', callbacks=None,
colsample_bylevel=1, colsample_bynode=1, colsample_bytree=0.7,
early_stopping_rounds=None, enable_categorical=False,
eval_metric=None, gamma=0, gpu_id=-1, grow_policy='depthwise',
importance_type=None, interaction_constraints='',
learning_rate=0.05, max_bin=256, max_cat_to_onehot=4,
max_delta_step=0, max_depth=6, max_leaves=0, min_child_weight=1,
missing=nan, monotone_constraints='()', n_estimators=100,
n_jobs=0, num_parallel_tree=1, predictor='auto', random_state=42,
reg_alpha=0, reg_lambda=1, ...)#xgb_classifier5(xgb_grid.best_estimator_
# training set
y_pred_xgb_tr55 = xgb_grid5.predict(X_train55)
accuracy_xgb_tr55 = accuracy_score(y_train55, y_pred_xgb_tr55)
print("XGBoost Model Score (Training Set): %.2f%%" % (accuracy_xgb_tr55 * 100.0))
# testing set
y_pred_xgb_t55 = xgb_grid5.predict(X_test55)
xgb_predictions55 = [round(value) for value in y_pred_xgb_t55]
# evaluate predictions
accuracy_xgb_t55 = accuracy_score(y_test55, xgb_predictions55)
print("XGBoost Model Score (Testing Set): %.2f%%" % (accuracy_xgb_t55 * 100.0))
XGBoost Model Score (Training Set): 88.31% XGBoost Model Score (Testing Set): 73.44%
plot_confusion_matrix(xgb_grid5, X_test55, y_test55,cmap="YlGnBu"); # good improvement
# Printing the classification report
print(classification_report(y_test55, y_pred_xgb_t55))
precision recall f1-score support
0 0.83 0.73 0.78 318
1 0.61 0.74 0.67 179
accuracy 0.73 497
macro avg 0.72 0.74 0.72 497
weighted avg 0.75 0.73 0.74 497
# Accuracy Score for each company.
xgb_accuracy1 = {'Aramco': accuracy_xgb_t11*100, 'SARCO': accuracy_xgb_t22*100, 'Rabigh': accuracy_xgb_t33*100, 'Bahri':accuracy_xgb_t44*100, 'Aldrees': accuracy_xgb_t55*100}
print(xgb_accuracy1)
xgb_names1 = list(xgb_accuracy1.keys())
xgb_values1 = list(xgb_accuracy1.values())
fig, ax = plt.subplots(figsize=(15,8))
plt.title('Optimized XGBoost Accuracy Score per Company (GridSearch CV)')
bars = ax.barh(xgb_names1, xgb_values1)
ax.bar_label(bars, fmt='%.2f%%')
plt.show()
{'Aramco': 62.68656716417911, 'SARCO': 94.16498993963782, 'Rabigh': 96.579476861167, 'Bahri': 66.39839034205231, 'Aldrees': 73.44064386317908}
df_xgb = pd.DataFrame({
'Company': xgb_names,
'XGBoost': xgb_values,
'XGBoost_GridSearch': xgb_values1,
'XGBoost_RanomizedSearch': xgb_values11
})
df_xgb
#df_xgb.to_csv('xgb_accuracy_1.csv')
| Company | XGBoost | XGBoost_GridSearch | XGBoost_RanomizedSearch | |
|---|---|---|---|---|
| 0 | Aramco | 66.417910 | 62.686567 | 67.910448 |
| 1 | SARCO | 93.561368 | 94.164990 | 94.567404 |
| 2 | Rabigh | 93.963783 | 96.579477 | 96.579477 |
| 3 | Bahri | 64.989940 | 66.398390 | 67.203219 |
| 4 | Aldrees | 72.837022 | 73.440644 | 73.843058 |
df_xgb.iloc[1,3]
# plt.figure(figsize=(15,6))
# plt.subplot(1, 3, 1)
# bar = sns.barplot(x="Company", y="XGBoost",data=df_xgb)
# for i in ax.containers:
# bar.bar_label(i, fmt='%.2f%%', label_type='center')
# plt.subplot(1, 3, 2)
# bar1 = sns.barplot (x="Company", y=df_xgb['XGBoost_GridSearch'].values, data=df_xgb);
# for i in ax.containers:
# bar1.bar_label(i, fmt='%.2f%%', label_type='center')
# plt.subplot(1, 3, 3)
# bar11 = sns.barplot(x=df_xgb['Company'], y=df_xgb['XGBoost_RanomizedSearch'].values, data=df_xgb);
# for i in ax.containers:
# bar11.bar_label(i, fmt='%.2f%%', label_type='center')
# plt.tight_layout()
# plt.suptitle('XGBoost Accuracy per Company', y=1.05)
# sns.despine(left=True, bottom=True)
# plt.show()
xgb_log_loss = {
'Aramco': log_loss(y_true=y_test1, y_pred=y_pred_xgb_t1, eps=1e-15, normalize=True, sample_weight=None, labels=None),
'SARCO': log_loss(y_true=y_test2, y_pred=y_pred_xgb_t2, eps=1e-15, normalize=True, sample_weight=None, labels=None),
'Rabigh': log_loss(y_true=y_test3, y_pred=y_pred_xgb_t3, eps=1e-15, normalize=True, sample_weight=None, labels=None),
'Bahri': log_loss(y_true=y_test4, y_pred=y_pred_xgb_t4, eps=1e-15, normalize=True, sample_weight=None, labels=None),
'Aldrees': log_loss(y_true=y_test5, y_pred=y_pred_xgb_t5, eps=1e-15, normalize=True, sample_weight=None, labels=None)
}
xgb_log_loss_values = list(xgb_log_loss.values())
xgb_randomsearch_log_loss = {
'Aramco': log_loss(y_true=y_test11, y_pred=y_pred_xgb_t111, eps=1e-15, normalize=True, sample_weight=None, labels=None),
'SARCO': log_loss(y_true=y_test22, y_pred=y_pred_xgb_t222, eps=1e-15, normalize=True, sample_weight=None, labels=None),
'Rabigh': log_loss(y_true=y_test33, y_pred=y_pred_xgb_t333, eps=1e-15, normalize=True, sample_weight=None, labels=None),
'Bahri': log_loss(y_true=y_test44, y_pred=y_pred_xgb_t444, eps=1e-15, normalize=True, sample_weight=None, labels=None),
'Aldrees': log_loss(y_true=y_test55, y_pred=y_pred_xgb_t555, eps=1e-15, normalize=True, sample_weight=None, labels=None)
}
xgb_randomsearch_log_loss_values = list(xgb_randomsearch_log_loss.values())
xgb_gridsearch_log_loss = {
'Aramco': log_loss(y_true=y_test11, y_pred=y_pred_xgb_t11, eps=1e-15, normalize=True, sample_weight=None, labels=None),
'SARCO': log_loss(y_true=y_test22, y_pred=y_pred_xgb_t22, eps=1e-15, normalize=True, sample_weight=None, labels=None),
'Rabigh': log_loss(y_true=y_test33, y_pred=y_pred_xgb_t33, eps=1e-15, normalize=True, sample_weight=None, labels=None),
'Bahri': log_loss(y_true=y_test44, y_pred=y_pred_xgb_t44, eps=1e-15, normalize=True, sample_weight=None, labels=None),
'Aldrees': log_loss(y_true=y_test55, y_pred=y_pred_xgb_t55, eps=1e-15, normalize=True, sample_weight=None, labels=None)
}
xgb_gridsearch_log_loss_values = list(xgb_gridsearch_log_loss.values())
df_xgb_log_loss = pd.DataFrame({
'Company': xgb_names,
'XGBoost': xgb_log_loss_values,
'XGBoost_GridSearch': xgb_gridsearch_log_loss_values,
'XGBoost_RanomizedSearch': xgb_randomsearch_log_loss_values
})
df_xgb_log_loss
#df_xgb_log_loss.to_csv('xgb_log_loss_1.csv')
| Company | XGBoost | XGBoost_GridSearch | XGBoost_RanomizedSearch | |
|---|---|---|---|---|
| 0 | Aramco | 11.598980 | 12.887740 | 11.083416 |
| 1 | SARCO | 2.223842 | 2.015352 | 1.876368 |
| 2 | Rabigh | 2.084861 | 1.181434 | 1.181434 |
| 3 | Bahri | 12.092196 | 11.605744 | 11.327737 |
| 4 | Aldrees | 9.381858 | 9.173413 | 9.034411 |
xgb_f1_score = {
'Aramco': f1_score(y_true=y_test1, y_pred=y_pred_xgb_t1),
'SARCO': f1_score(y_true=y_test2, y_pred=y_pred_xgb_t2),
'Rabigh': f1_score(y_true=y_test3, y_pred=y_pred_xgb_t3),
'Bahri': f1_score(y_true=y_test4, y_pred=y_pred_xgb_t4),
'Aldrees': f1_score(y_true=y_test5, y_pred=y_pred_xgb_t5,)
}
xgb_f1_score_values = list(xgb_f1_score.values())
xgb_randomsearch_f1_score = {
'Aramco': f1_score(y_true=y_test11, y_pred=y_pred_xgb_t111),
'SARCO': f1_score(y_true=y_test22, y_pred=y_pred_xgb_t222),
'Rabigh': f1_score(y_true=y_test33, y_pred=y_pred_xgb_t333),
'Bahri': f1_score(y_true=y_test44, y_pred=y_pred_xgb_t444),
'Aldrees': f1_score(y_true=y_test55, y_pred=y_pred_xgb_t555)
}
xgb_randomsearch_f1_score_values = list(xgb_randomsearch_f1_score.values())
xgb_gridsearch_f1_score = {
'Aramco': f1_score(y_true=y_test11, y_pred=y_pred_xgb_t11),
'SARCO': f1_score(y_true=y_test22, y_pred=y_pred_xgb_t22),
'Rabigh': f1_score(y_true=y_test33, y_pred=y_pred_xgb_t33),
'Bahri': f1_score(y_true=y_test44, y_pred=y_pred_xgb_t44),
'Aldrees': f1_score(y_true=y_test55, y_pred=y_pred_xgb_t55)
}
xgb_gridsearch_f1_score_values = list(xgb_gridsearch_f1_score.values())
df_xgb_f1_score = pd.DataFrame({
'Company': xgb_names,
'XGBoost': xgb_f1_score_values,
'XGBoost_GridSearch': xgb_gridsearch_f1_score_values,
'XGBoost_RanomizedSearch': xgb_randomsearch_f1_score_values
})
df_xgb_f1_score
#df_xgb_f1_score.to_csv('xgb_f1_score_1.csv')
| Company | XGBoost | XGBoost_GridSearch | XGBoost_RanomizedSearch | |
|---|---|---|---|---|
| 0 | Aramco | 0.720497 | 0.679487 | 0.699301 |
| 1 | SARCO | 0.333333 | 0.325581 | 0.470588 |
| 2 | Rabigh | 0.968815 | 0.982600 | 0.982600 |
| 3 | Bahri | 0.690391 | 0.712565 | 0.703097 |
| 4 | Aldrees | 0.608696 | 0.666667 | 0.659686 |
# split data
X_train1, X_test1, y_train1, y_test1 = split_data(df_Aramco, 'Target')
#scaling the data
X_train1, X_test1 = scale_data(X_train1, X_test1)
# split data
X_train2, X_test2, y_train2, y_test2 = split_data(df_SARCO, 'Target')
#scaling the data
X_train2, X_test2 = scale_data(X_train2, X_test2)
# split data
X_train3, X_test3, y_train3, y_test3 = split_data(df_Rabigh, 'Target')
#scaling the data
X_train3, X_test3 = scale_data(X_train3, X_test3)
X_train4, X_test4, y_train4, y_test4 = split_data(df_Bahri, 'Target')
#scaling the data
X_train4, X_test4 = scale_data(X_train4, X_test4)
X_train5, X_test5, y_train5, y_test5 = split_data(df_Aldrees, 'Target')
#scaling the data
X_train5, X_test5 = scale_data(X_train5, X_test5)
svc = SVC(kernel = 'linear', random_state=42,class_weight="balanced")
svc.fit(X_train1, y_train1)
SVC(class_weight='balanced', kernel='linear', random_state=42)In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook.
SVC(class_weight='balanced', kernel='linear', random_state=42)
# training set
y_pred_svc_tr1 = svc.predict(X_train1)
accuracy_svc_tr1 = accuracy_score(y_train1, y_pred_svc_tr1)
print("Support Vector Classification Model Score (Training Set): %.2f%%" % (accuracy_svc_tr1 * 100.0))
# testing set
y_pred_svc_t1 = svc.predict(X_test1)
predictions_svc1 = [round(value) for value in y_pred_svc_t1]
# evaluate predictions
accuracy_svc_t1 = accuracy_score(y_test1, predictions_svc1)
print("Support Vector Classification Model Score (Testing Set): %.2f%%" % (accuracy_svc_t1 * 100.0))
Support Vector Classification Model Score (Training Set): 64.18% Support Vector Classification Model Score (Testing Set): 67.91%
plot_confusion_matrix(svc, X_test1, y_test1,cmap="YlGnBu");
# Printing the classification report
print(classification_report(y_test1, y_pred_svc_t1))
precision recall f1-score support
0 0.60 0.61 0.61 54
1 0.73 0.72 0.73 80
accuracy 0.68 134
macro avg 0.67 0.67 0.67 134
weighted avg 0.68 0.68 0.68 134
svc.fit(X_train2, y_train2)
SVC(class_weight='balanced', kernel='linear', random_state=42)In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook.
SVC(class_weight='balanced', kernel='linear', random_state=42)
# training set
y_pred_svc_tr2 = svc.predict(X_train2)
accuracy_svc_tr2 = accuracy_score(y_train2, y_pred_svc_tr2)
print("Support Vector Classification Model Score (Training Set): %.2f%%" % (accuracy_svc_tr2 * 100.0))
# testing set
y_pred_svc_t2 = svc.predict(X_test2)
predictions_svc2 = [round(value) for value in y_pred_svc_t2]
# evaluate predictions
accuracy_svc_t2 = accuracy_score(y_test2, predictions_svc2)
print("Support Vector Classification Model Score (Testing Set): %.2f%%" % (accuracy_svc_t2 * 100.0))
Support Vector Classification Model Score (Training Set): 70.51% Support Vector Classification Model Score (Testing Set): 71.43%
plot_confusion_matrix(svc, X_test2, y_test2,cmap="YlGnBu");
# Printing the classification report
print(classification_report(y_test2, y_pred_svc_t2))
precision recall f1-score support
0 0.98 0.71 0.82 468
1 0.14 0.76 0.24 29
accuracy 0.71 497
macro avg 0.56 0.74 0.53 497
weighted avg 0.93 0.71 0.79 497
svc.fit(X_train3, y_train3)
SVC(class_weight='balanced', kernel='linear', random_state=42)In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook.
SVC(class_weight='balanced', kernel='linear', random_state=42)
# training set
y_pred_svc_tr3 = svc.predict(X_train3)
accuracy_svc_tr3 = accuracy_score(y_train3, y_pred_svc_tr3)
print("Support Vector Classification Model Score (Training Set): %.2f%%" % (accuracy_svc_tr3 * 100.0))
# testing set
y_pred_svc_t3 = svc.predict(X_test3)
predictions_svc3 = [round(value) for value in y_pred_svc_t3]
# evaluate predictions
accuracy_svc_t3 = accuracy_score(y_test3, predictions_svc3)
print("Support Vector Classification Model Score (Testing Set): %.2f%%" % (accuracy_svc_t3 * 100.0))
Support Vector Classification Model Score (Training Set): 59.14% Support Vector Classification Model Score (Testing Set): 56.74%
plot_confusion_matrix(svc, X_test3, y_test3,cmap="YlGnBu")
<sklearn.metrics._plot.confusion_matrix.ConfusionMatrixDisplay at 0x1d6ae7588e0>
# Printing the classification report
print(classification_report(y_test3, y_pred_svc_t3))
precision recall f1-score support
0 0.05 0.59 0.09 17
1 0.97 0.57 0.72 480
accuracy 0.57 497
macro avg 0.51 0.58 0.40 497
weighted avg 0.94 0.57 0.70 497
svc.fit(X_train4, y_train4)
SVC(class_weight='balanced', kernel='linear', random_state=42)In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook.
SVC(class_weight='balanced', kernel='linear', random_state=42)
# training set
y_pred_svc_tr4 = svc.predict(X_train4)
accuracy_svc_tr4 = accuracy_score(y_train4, y_pred_svc_tr4)
print("Support Vector Classification Model Score (Training Set): %.2f%%" % (accuracy_svc_tr4 * 100.0))
# testing set
y_pred_svc_t4 = svc.predict(X_test4)
predictions_svc4 = [round(value) for value in y_pred_svc_t4]
# evaluate predictions
accuracy_svc_t4 = accuracy_score(y_test4, predictions_svc4)
print("Support Vector Classification Model Score (Testing Set): %.2f%%" % (accuracy_svc_t4 * 100.0))
Support Vector Classification Model Score (Training Set): 65.62% Support Vector Classification Model Score (Testing Set): 62.37%
plot_confusion_matrix(svc, X_test4, y_test4,cmap="YlGnBu")
<sklearn.metrics._plot.confusion_matrix.ConfusionMatrixDisplay at 0x1d6aab23250>
# Printing the classification report
print(classification_report(y_test4, y_pred_svc_t4))
precision recall f1-score support
0 0.56 0.71 0.63 222
1 0.70 0.55 0.62 275
accuracy 0.62 497
macro avg 0.63 0.63 0.62 497
weighted avg 0.64 0.62 0.62 497
svc.fit(X_train5, y_train5)
SVC(class_weight='balanced', kernel='linear', random_state=42)In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook.
SVC(class_weight='balanced', kernel='linear', random_state=42)
# training set
y_pred_svc_tr5 = svc.predict(X_train5)
accuracy_svc_tr5 = accuracy_score(y_train5, y_pred_svc_tr5)
print("Support Vector Classification Model Score (Training Set): %.2f%%" % (accuracy_svc_tr5 * 100.0))
# testing set
y_pred_svc_t5 = svc.predict(X_test5)
predictions_svc5 = [round(value) for value in y_pred_svc_t5]
# evaluate predictions
accuracy_svc_t5 = accuracy_score(y_test5, predictions_svc5)
print("Support Vector Classification Model Score (Testing Set): %.2f%%" % (accuracy_svc_t5 * 100.0))
Support Vector Classification Model Score (Training Set): 64.94% Support Vector Classification Model Score (Testing Set): 66.00%
plot_confusion_matrix(svc, X_test5, y_test5,cmap="YlGnBu");
# Printing the classification report
print(classification_report(y_test5, y_pred_svc_t5))
precision recall f1-score support
0 0.81 0.62 0.70 318
1 0.52 0.74 0.61 179
accuracy 0.66 497
macro avg 0.66 0.68 0.65 497
weighted avg 0.70 0.66 0.67 497
# Accuracy Score for each company.
svc_accuracy = {'Aramco': accuracy_svc_t1*100, 'SARCO': accuracy_svc_t2*100, 'Rabigh': accuracy_svc_t3*100, 'Bahri':accuracy_svc_t4*100, 'Aldrees': accuracy_svc_t5*100}
print(svc_accuracy)
svc_names = list(svc_accuracy.keys())
svc_values = list(svc_accuracy.values())
fig, ax = plt.subplots(figsize=(15,8))
plt.title('Support Vector Classification Accuracy Score per Company')
svc_bars = ax.barh(svc_names, svc_values)
ax.bar_label(svc_bars)
plt.show()
{'Aramco': 67.91044776119402, 'SARCO': 71.42857142857143, 'Rabigh': 56.74044265593562, 'Bahri': 62.37424547283702, 'Aldrees': 65.99597585513078}
auc = make_scorer(roc_auc_score)
rand_list = {"C": stats.uniform(2, 10),
"gamma": stats.uniform(0.1, 1)}
svc_random_search = RandomizedSearchCV(svc, param_distributions = rand_list, n_iter = 20, n_jobs = 4, cv = 3, random_state = 2017, scoring = auc)
# split data
X_train1, X_test1, y_train1, y_test1 = split_data(df_Aramco.drop('SMA_30', axis=1), 'Target')
#scaling the data
X_train1, X_test1 = scale_data(X_train1, X_test1)
# split data
X_train2, X_test2, y_train2, y_test2 = split_data(df_SARCO.drop('SMA_30', axis=1), 'Target')
#scaling the data
X_train2, X_test2 = scale_data(X_train2, X_test2)
# split data
X_train3, X_test3, y_train3, y_test3 = split_data(df_Rabigh.drop('SMA_30', axis=1), 'Target')
#scaling the data
X_train3, X_test3 = scale_data(X_train3, X_test3)
X_train4, X_test4, y_train4, y_test4 = split_data(df_Bahri.drop('SMA_30', axis=1), 'Target')
#scaling the data
X_train4, X_test4 = scale_data(X_train4, X_test4)
X_train5, X_test5, y_train5, y_test5 = split_data(df_Aldrees.drop('SMA_30', axis=1), 'Target')
#scaling the data
X_train5, X_test5 = scale_data(X_train5, X_test5)
# Fit the random search model
svc_random_search.fit(X_train1, y_train1)
RandomizedSearchCV(cv=3,
estimator=SVC(class_weight='balanced', kernel='linear',
random_state=42),
n_iter=20, n_jobs=4,
param_distributions={'C': <scipy.stats._distn_infrastructure.rv_frozen object at 0x000001D6B24E1F40>,
'gamma': <scipy.stats._distn_infrastructure.rv_frozen object at 0x000001D6B2566340>},
random_state=2017, scoring=make_scorer(roc_auc_score))In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. RandomizedSearchCV(cv=3,
estimator=SVC(class_weight='balanced', kernel='linear',
random_state=42),
n_iter=20, n_jobs=4,
param_distributions={'C': <scipy.stats._distn_infrastructure.rv_frozen object at 0x000001D6B24E1F40>,
'gamma': <scipy.stats._distn_infrastructure.rv_frozen object at 0x000001D6B2566340>},
random_state=2017, scoring=make_scorer(roc_auc_score))SVC(class_weight='balanced', kernel='linear', random_state=42)
SVC(class_weight='balanced', kernel='linear', random_state=42)
# training set
y_pred_svc_tr11 = svc_random_search.predict(X_train1)
accuracy_svc_tr11 = accuracy_score(y_train1, y_pred_svc_tr11)
print("Support Vector Classification Model Score (Training Set): %.2f%%" % (accuracy_svc_tr11 * 100.0))
# testing set
y_pred_svc_t11 = svc_random_search.predict(X_test1)
predictions_svc11 = [round(value) for value in y_pred_svc_t11]
# evaluate predictions
accuracy_svc_t11 = accuracy_score(y_test1, predictions_svc11)
print("Support Vector Classification Model Score (Testing Set): %.2f%%" % (accuracy_svc_t11 * 100.0))
Support Vector Classification Model Score (Training Set): 64.37% Support Vector Classification Model Score (Testing Set): 64.93%
plot_confusion_matrix(svc_random_search, X_test1, y_test1,cmap="YlGnBu");
# Printing the classification report
print(classification_report(y_test1, y_pred_svc_t11))
precision recall f1-score support
0 0.56 0.57 0.57 54
1 0.71 0.70 0.70 80
accuracy 0.65 134
macro avg 0.64 0.64 0.64 134
weighted avg 0.65 0.65 0.65 134
svc_random_search.fit(X_train2, y_train2)
RandomizedSearchCV(cv=3,
estimator=SVC(class_weight='balanced', kernel='linear',
random_state=42),
n_iter=20, n_jobs=4,
param_distributions={'C': <scipy.stats._distn_infrastructure.rv_frozen object at 0x000001D6B24E1F40>,
'gamma': <scipy.stats._distn_infrastructure.rv_frozen object at 0x000001D6B2566340>},
random_state=2017, scoring=make_scorer(roc_auc_score))In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. RandomizedSearchCV(cv=3,
estimator=SVC(class_weight='balanced', kernel='linear',
random_state=42),
n_iter=20, n_jobs=4,
param_distributions={'C': <scipy.stats._distn_infrastructure.rv_frozen object at 0x000001D6B24E1F40>,
'gamma': <scipy.stats._distn_infrastructure.rv_frozen object at 0x000001D6B2566340>},
random_state=2017, scoring=make_scorer(roc_auc_score))SVC(class_weight='balanced', kernel='linear', random_state=42)
SVC(class_weight='balanced', kernel='linear', random_state=42)
# training set
y_pred_svc_tr22 = svc_random_search.predict(X_train2)
accuracy_svc_tr22 = accuracy_score(y_train2, y_pred_svc_tr22)
print("Support Vector Classification Model Score (Training Set): %.2f%%" % (accuracy_svc_tr22 * 100.0))
# testing set
y_pred_svc_t22 = svc_random_search.predict(X_test2)
predictions_svc22 = [round(value) for value in y_pred_svc_t22]
# evaluate predictions
accuracy_svc_t22 = accuracy_score(y_test2, predictions_svc22)
print("Support Vector Classification Model Score (Testing Set): %.2f%%" % (accuracy_svc_t22 * 100.0))
Support Vector Classification Model Score (Training Set): 70.56% Support Vector Classification Model Score (Testing Set): 71.83%
plot_confusion_matrix(svc_random_search, X_test2, y_test2,cmap="YlGnBu");
# Printing the classification report
print(classification_report(y_test2, y_pred_svc_t22))
precision recall f1-score support
0 0.98 0.72 0.83 468
1 0.14 0.76 0.24 29
accuracy 0.72 497
macro avg 0.56 0.74 0.53 497
weighted avg 0.93 0.72 0.79 497
svc_random_search.fit(X_train3, y_train3)
RandomizedSearchCV(cv=3,
estimator=SVC(class_weight='balanced', kernel='linear',
random_state=42),
n_iter=20, n_jobs=4,
param_distributions={'C': <scipy.stats._distn_infrastructure.rv_frozen object at 0x000001D6B24E1F40>,
'gamma': <scipy.stats._distn_infrastructure.rv_frozen object at 0x000001D6B2566340>},
random_state=2017, scoring=make_scorer(roc_auc_score))In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. RandomizedSearchCV(cv=3,
estimator=SVC(class_weight='balanced', kernel='linear',
random_state=42),
n_iter=20, n_jobs=4,
param_distributions={'C': <scipy.stats._distn_infrastructure.rv_frozen object at 0x000001D6B24E1F40>,
'gamma': <scipy.stats._distn_infrastructure.rv_frozen object at 0x000001D6B2566340>},
random_state=2017, scoring=make_scorer(roc_auc_score))SVC(class_weight='balanced', kernel='linear', random_state=42)
SVC(class_weight='balanced', kernel='linear', random_state=42)
# training set
y_pred_svc_tr33 = svc_random_search.predict(X_train3)
accuracy_svc_tr33 = accuracy_score(y_train3, y_pred_svc_tr33)
print("Support Vector Classification Model Score (Training Set): %.2f%%" % (accuracy_svc_tr33 * 100.0))
# testing set
y_pred_svc_t33 = svc_random_search.predict(X_test3)
predictions_svc33 = [round(value) for value in y_pred_svc_t33]
# evaluate predictions
accuracy_svc_t33 = accuracy_score(y_test3, predictions_svc33)
print("Support Vector Classification Model Score (Testing Set): %.2f%%" % (accuracy_svc_t33 * 100.0))
Support Vector Classification Model Score (Training Set): 58.39% Support Vector Classification Model Score (Testing Set): 56.14%
plot_confusion_matrix(svc_random_search, X_test3, y_test3,cmap="YlGnBu");
# Printing the classification report
print(classification_report(y_test3, y_pred_svc_t33))
precision recall f1-score support
0 0.05 0.59 0.08 17
1 0.97 0.56 0.71 480
accuracy 0.56 497
macro avg 0.51 0.57 0.40 497
weighted avg 0.94 0.56 0.69 497
svc_random_search.fit(X_train4, y_train4)
RandomizedSearchCV(cv=3,
estimator=SVC(class_weight='balanced', kernel='linear',
random_state=42),
n_iter=20, n_jobs=4,
param_distributions={'C': <scipy.stats._distn_infrastructure.rv_frozen object at 0x000001D6B24E1F40>,
'gamma': <scipy.stats._distn_infrastructure.rv_frozen object at 0x000001D6B2566340>},
random_state=2017, scoring=make_scorer(roc_auc_score))In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. RandomizedSearchCV(cv=3,
estimator=SVC(class_weight='balanced', kernel='linear',
random_state=42),
n_iter=20, n_jobs=4,
param_distributions={'C': <scipy.stats._distn_infrastructure.rv_frozen object at 0x000001D6B24E1F40>,
'gamma': <scipy.stats._distn_infrastructure.rv_frozen object at 0x000001D6B2566340>},
random_state=2017, scoring=make_scorer(roc_auc_score))SVC(class_weight='balanced', kernel='linear', random_state=42)
SVC(class_weight='balanced', kernel='linear', random_state=42)
# training set
y_pred_svc_tr44 = svc_random_search.predict(X_train4)
accuracy_svc_tr44 = accuracy_score(y_train4, y_pred_svc_tr44)
print("Support Vector Classification Model Score (Training Set): %.2f%%" % (accuracy_svc_tr44 * 100.0))
# testing set
y_pred_svc_t44 = svc_random_search.predict(X_test4)
predictions_svc44 = [round(value) for value in y_pred_svc_t44]
# evaluate predictions
accuracy_svc_t44 = accuracy_score(y_test4, predictions_svc44)
print("Support Vector Classification Model Score (Testing Set): %.2f%%" % (accuracy_svc_t44 * 100.0))
Support Vector Classification Model Score (Training Set): 65.78% Support Vector Classification Model Score (Testing Set): 62.37%
plot_confusion_matrix(svc_random_search, X_test4, y_test4,cmap="YlGnBu");
# Printing the classification report
print(classification_report(y_test4, y_pred_svc_t44))
precision recall f1-score support
0 0.56 0.71 0.63 222
1 0.70 0.55 0.62 275
accuracy 0.62 497
macro avg 0.63 0.63 0.62 497
weighted avg 0.64 0.62 0.62 497
svc_random_search.fit(X_train5, y_train5)
RandomizedSearchCV(cv=3,
estimator=SVC(class_weight='balanced', kernel='linear',
random_state=42),
n_iter=20, n_jobs=4,
param_distributions={'C': <scipy.stats._distn_infrastructure.rv_frozen object at 0x000001D6B24E1F40>,
'gamma': <scipy.stats._distn_infrastructure.rv_frozen object at 0x000001D6B2566340>},
random_state=2017, scoring=make_scorer(roc_auc_score))In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. RandomizedSearchCV(cv=3,
estimator=SVC(class_weight='balanced', kernel='linear',
random_state=42),
n_iter=20, n_jobs=4,
param_distributions={'C': <scipy.stats._distn_infrastructure.rv_frozen object at 0x000001D6B24E1F40>,
'gamma': <scipy.stats._distn_infrastructure.rv_frozen object at 0x000001D6B2566340>},
random_state=2017, scoring=make_scorer(roc_auc_score))SVC(class_weight='balanced', kernel='linear', random_state=42)
SVC(class_weight='balanced', kernel='linear', random_state=42)
# training set
y_pred_svc_tr55 = svc_random_search.predict(X_train5)
accuracy_svc_tr55 = accuracy_score(y_train5, y_pred_svc_tr55)
print("Support Vector Classification Model Score (Training Set): %.2f%%" % (accuracy_svc_tr55 * 100.0))
# testing set
y_pred_svc_t55 = svc_random_search.predict(X_test5)
predictions_svc55 = [round(value) for value in y_pred_svc_t55]
# evaluate predictions
accuracy_svc_t55 = accuracy_score(y_test5, predictions_svc55)
print("Support Vector Classification Model Score (Testing Set): %.2f%%" % (accuracy_svc_t55 * 100.0))
Support Vector Classification Model Score (Training Set): 64.63% Support Vector Classification Model Score (Testing Set): 66.80%
plot_confusion_matrix(svc_random_search, X_test5, y_test5,cmap="YlGnBu")
<sklearn.metrics._plot.confusion_matrix.ConfusionMatrixDisplay at 0x1d6b3439d30>
# Printing the classification report
print(classification_report(y_test5, y_pred_svc_t55))
precision recall f1-score support
0 0.81 0.63 0.71 318
1 0.53 0.74 0.62 179
accuracy 0.67 497
macro avg 0.67 0.68 0.66 497
weighted avg 0.71 0.67 0.67 497
# Accuracy Score for each company.
svc_accuracy2 = {'Aramco': accuracy_svc_t11*100, 'SARCO': accuracy_svc_t22*100, 'Rabigh': accuracy_svc_t33*100, 'Bahri':accuracy_svc_t44*100, 'Aldrees': accuracy_svc_t55*100}
svc_names2 = list(svc_accuracy2.keys())
svc_values2 = list(svc_accuracy2.values())
fig, ax = plt.subplots(figsize=(15,8))
plt.title('Support Vector Classification using RandomSearch CV Classification Accuracy Score per Company')
svc_bars2 = ax.barh(svc_names2, svc_values2)
ax.bar_label(svc_bars2, padding=5)
plt.xlim([0,100])
plt.show()
# defining parameter range
param_grid = {'C': [0.1,1, 10, 100],
'gamma': [1,0.1,0.01,0.001],
'kernel': ['rbf', 'sigmoid']}
SVC_grid = GridSearchCV(SVC(class_weight="balanced"), param_grid, refit = True, verbose = 3)
SVC_grid.fit(X_train1, y_train1)
Fitting 5 folds for each of 32 candidates, totalling 160 fits [CV 1/5] END ........C=0.1, gamma=1, kernel=rbf;, score=0.537 total time= 0.0s [CV 2/5] END ........C=0.1, gamma=1, kernel=rbf;, score=0.570 total time= 0.0s [CV 3/5] END ........C=0.1, gamma=1, kernel=rbf;, score=0.551 total time= 0.0s [CV 4/5] END ........C=0.1, gamma=1, kernel=rbf;, score=0.570 total time= 0.0s [CV 5/5] END ........C=0.1, gamma=1, kernel=rbf;, score=0.551 total time= 0.0s [CV 1/5] END ....C=0.1, gamma=1, kernel=sigmoid;, score=0.611 total time= 0.0s [CV 2/5] END ....C=0.1, gamma=1, kernel=sigmoid;, score=0.542 total time= 0.0s [CV 3/5] END ....C=0.1, gamma=1, kernel=sigmoid;, score=0.430 total time= 0.0s [CV 4/5] END ....C=0.1, gamma=1, kernel=sigmoid;, score=0.551 total time= 0.0s [CV 5/5] END ....C=0.1, gamma=1, kernel=sigmoid;, score=0.505 total time= 0.0s [CV 1/5] END ......C=0.1, gamma=0.1, kernel=rbf;, score=0.630 total time= 0.0s [CV 2/5] END ......C=0.1, gamma=0.1, kernel=rbf;, score=0.617 total time= 0.0s [CV 3/5] END ......C=0.1, gamma=0.1, kernel=rbf;, score=0.607 total time= 0.0s [CV 4/5] END ......C=0.1, gamma=0.1, kernel=rbf;, score=0.551 total time= 0.0s [CV 5/5] END ......C=0.1, gamma=0.1, kernel=rbf;, score=0.486 total time= 0.0s [CV 1/5] END ..C=0.1, gamma=0.1, kernel=sigmoid;, score=0.731 total time= 0.0s [CV 2/5] END ..C=0.1, gamma=0.1, kernel=sigmoid;, score=0.626 total time= 0.0s [CV 3/5] END ..C=0.1, gamma=0.1, kernel=sigmoid;, score=0.607 total time= 0.0s [CV 4/5] END ..C=0.1, gamma=0.1, kernel=sigmoid;, score=0.598 total time= 0.0s [CV 5/5] END ..C=0.1, gamma=0.1, kernel=sigmoid;, score=0.570 total time= 0.0s [CV 1/5] END .....C=0.1, gamma=0.01, kernel=rbf;, score=0.713 total time= 0.0s [CV 2/5] END .....C=0.1, gamma=0.01, kernel=rbf;, score=0.654 total time= 0.0s [CV 3/5] END .....C=0.1, gamma=0.01, kernel=rbf;, score=0.626 total time= 0.0s [CV 4/5] END .....C=0.1, gamma=0.01, kernel=rbf;, score=0.579 total time= 0.0s [CV 5/5] END .....C=0.1, gamma=0.01, kernel=rbf;, score=0.589 total time= 0.0s [CV 1/5] END .C=0.1, gamma=0.01, kernel=sigmoid;, score=0.472 total time= 0.0s [CV 2/5] END .C=0.1, gamma=0.01, kernel=sigmoid;, score=0.607 total time= 0.0s [CV 3/5] END .C=0.1, gamma=0.01, kernel=sigmoid;, score=0.570 total time= 0.0s [CV 4/5] END .C=0.1, gamma=0.01, kernel=sigmoid;, score=0.579 total time= 0.0s [CV 5/5] END .C=0.1, gamma=0.01, kernel=sigmoid;, score=0.561 total time= 0.0s [CV 1/5] END ....C=0.1, gamma=0.001, kernel=rbf;, score=0.444 total time= 0.0s [CV 2/5] END ....C=0.1, gamma=0.001, kernel=rbf;, score=0.561 total time= 0.0s [CV 3/5] END ....C=0.1, gamma=0.001, kernel=rbf;, score=0.561 total time= 0.0s [CV 4/5] END ....C=0.1, gamma=0.001, kernel=rbf;, score=0.561 total time= 0.0s [CV 5/5] END ....C=0.1, gamma=0.001, kernel=rbf;, score=0.561 total time= 0.0s [CV 1/5] END C=0.1, gamma=0.001, kernel=sigmoid;, score=0.444 total time= 0.0s [CV 2/5] END C=0.1, gamma=0.001, kernel=sigmoid;, score=0.561 total time= 0.0s [CV 3/5] END C=0.1, gamma=0.001, kernel=sigmoid;, score=0.561 total time= 0.0s [CV 4/5] END C=0.1, gamma=0.001, kernel=sigmoid;, score=0.561 total time= 0.0s [CV 5/5] END C=0.1, gamma=0.001, kernel=sigmoid;, score=0.561 total time= 0.0s [CV 1/5] END ..........C=1, gamma=1, kernel=rbf;, score=0.528 total time= 0.0s [CV 2/5] END ..........C=1, gamma=1, kernel=rbf;, score=0.589 total time= 0.0s [CV 3/5] END ..........C=1, gamma=1, kernel=rbf;, score=0.495 total time= 0.0s [CV 4/5] END ..........C=1, gamma=1, kernel=rbf;, score=0.505 total time= 0.0s [CV 5/5] END ..........C=1, gamma=1, kernel=rbf;, score=0.533 total time= 0.0s [CV 1/5] END ......C=1, gamma=1, kernel=sigmoid;, score=0.593 total time= 0.0s [CV 2/5] END ......C=1, gamma=1, kernel=sigmoid;, score=0.533 total time= 0.0s [CV 3/5] END ......C=1, gamma=1, kernel=sigmoid;, score=0.598 total time= 0.0s [CV 4/5] END ......C=1, gamma=1, kernel=sigmoid;, score=0.570 total time= 0.0s [CV 5/5] END ......C=1, gamma=1, kernel=sigmoid;, score=0.486 total time= 0.0s [CV 1/5] END ........C=1, gamma=0.1, kernel=rbf;, score=0.611 total time= 0.0s [CV 2/5] END ........C=1, gamma=0.1, kernel=rbf;, score=0.626 total time= 0.0s [CV 3/5] END ........C=1, gamma=0.1, kernel=rbf;, score=0.607 total time= 0.0s [CV 4/5] END ........C=1, gamma=0.1, kernel=rbf;, score=0.458 total time= 0.0s [CV 5/5] END ........C=1, gamma=0.1, kernel=rbf;, score=0.505 total time= 0.0s [CV 1/5] END ....C=1, gamma=0.1, kernel=sigmoid;, score=0.519 total time= 0.0s [CV 2/5] END ....C=1, gamma=0.1, kernel=sigmoid;, score=0.561 total time= 0.0s [CV 3/5] END ....C=1, gamma=0.1, kernel=sigmoid;, score=0.542 total time= 0.0s [CV 4/5] END ....C=1, gamma=0.1, kernel=sigmoid;, score=0.467 total time= 0.0s [CV 5/5] END ....C=1, gamma=0.1, kernel=sigmoid;, score=0.477 total time= 0.0s [CV 1/5] END .......C=1, gamma=0.01, kernel=rbf;, score=0.639 total time= 0.0s [CV 2/5] END .......C=1, gamma=0.01, kernel=rbf;, score=0.654 total time= 0.0s [CV 3/5] END .......C=1, gamma=0.01, kernel=rbf;, score=0.626 total time= 0.0s [CV 4/5] END .......C=1, gamma=0.01, kernel=rbf;, score=0.533 total time= 0.0s [CV 5/5] END .......C=1, gamma=0.01, kernel=rbf;, score=0.626 total time= 0.0s [CV 1/5] END ...C=1, gamma=0.01, kernel=sigmoid;, score=0.713 total time= 0.0s [CV 2/5] END ...C=1, gamma=0.01, kernel=sigmoid;, score=0.636 total time= 0.0s [CV 3/5] END ...C=1, gamma=0.01, kernel=sigmoid;, score=0.617 total time= 0.0s [CV 4/5] END ...C=1, gamma=0.01, kernel=sigmoid;, score=0.579 total time= 0.0s [CV 5/5] END ...C=1, gamma=0.01, kernel=sigmoid;, score=0.579 total time= 0.0s [CV 1/5] END ......C=1, gamma=0.001, kernel=rbf;, score=0.620 total time= 0.0s [CV 2/5] END ......C=1, gamma=0.001, kernel=rbf;, score=0.645 total time= 0.0s [CV 3/5] END ......C=1, gamma=0.001, kernel=rbf;, score=0.626 total time= 0.0s [CV 4/5] END ......C=1, gamma=0.001, kernel=rbf;, score=0.579 total time= 0.0s [CV 5/5] END ......C=1, gamma=0.001, kernel=rbf;, score=0.607 total time= 0.0s [CV 1/5] END ..C=1, gamma=0.001, kernel=sigmoid;, score=0.583 total time= 0.0s [CV 2/5] END ..C=1, gamma=0.001, kernel=sigmoid;, score=0.607 total time= 0.0s [CV 3/5] END ..C=1, gamma=0.001, kernel=sigmoid;, score=0.626 total time= 0.0s [CV 4/5] END ..C=1, gamma=0.001, kernel=sigmoid;, score=0.570 total time= 0.0s [CV 5/5] END ..C=1, gamma=0.001, kernel=sigmoid;, score=0.579 total time= 0.0s [CV 1/5] END .........C=10, gamma=1, kernel=rbf;, score=0.528 total time= 0.0s [CV 2/5] END .........C=10, gamma=1, kernel=rbf;, score=0.579 total time= 0.0s [CV 3/5] END .........C=10, gamma=1, kernel=rbf;, score=0.477 total time= 0.0s [CV 4/5] END .........C=10, gamma=1, kernel=rbf;, score=0.542 total time= 0.0s [CV 5/5] END .........C=10, gamma=1, kernel=rbf;, score=0.551 total time= 0.0s [CV 1/5] END .....C=10, gamma=1, kernel=sigmoid;, score=0.593 total time= 0.0s [CV 2/5] END .....C=10, gamma=1, kernel=sigmoid;, score=0.542 total time= 0.0s [CV 3/5] END .....C=10, gamma=1, kernel=sigmoid;, score=0.607 total time= 0.0s [CV 4/5] END .....C=10, gamma=1, kernel=sigmoid;, score=0.570 total time= 0.0s [CV 5/5] END .....C=10, gamma=1, kernel=sigmoid;, score=0.495 total time= 0.0s [CV 1/5] END .......C=10, gamma=0.1, kernel=rbf;, score=0.620 total time= 0.0s [CV 2/5] END .......C=10, gamma=0.1, kernel=rbf;, score=0.579 total time= 0.0s [CV 3/5] END .......C=10, gamma=0.1, kernel=rbf;, score=0.551 total time= 0.0s [CV 4/5] END .......C=10, gamma=0.1, kernel=rbf;, score=0.561 total time= 0.0s [CV 5/5] END .......C=10, gamma=0.1, kernel=rbf;, score=0.551 total time= 0.0s [CV 1/5] END ...C=10, gamma=0.1, kernel=sigmoid;, score=0.528 total time= 0.0s [CV 2/5] END ...C=10, gamma=0.1, kernel=sigmoid;, score=0.570 total time= 0.0s [CV 3/5] END ...C=10, gamma=0.1, kernel=sigmoid;, score=0.551 total time= 0.0s [CV 4/5] END ...C=10, gamma=0.1, kernel=sigmoid;, score=0.439 total time= 0.0s [CV 5/5] END ...C=10, gamma=0.1, kernel=sigmoid;, score=0.495 total time= 0.0s [CV 1/5] END ......C=10, gamma=0.01, kernel=rbf;, score=0.657 total time= 0.0s [CV 2/5] END ......C=10, gamma=0.01, kernel=rbf;, score=0.626 total time= 0.0s [CV 3/5] END ......C=10, gamma=0.01, kernel=rbf;, score=0.626 total time= 0.0s [CV 4/5] END ......C=10, gamma=0.01, kernel=rbf;, score=0.533 total time= 0.0s [CV 5/5] END ......C=10, gamma=0.01, kernel=rbf;, score=0.533 total time= 0.0s [CV 1/5] END ..C=10, gamma=0.01, kernel=sigmoid;, score=0.667 total time= 0.0s [CV 2/5] END ..C=10, gamma=0.01, kernel=sigmoid;, score=0.598 total time= 0.0s [CV 3/5] END ..C=10, gamma=0.01, kernel=sigmoid;, score=0.645 total time= 0.0s [CV 4/5] END ..C=10, gamma=0.01, kernel=sigmoid;, score=0.495 total time= 0.0s [CV 5/5] END ..C=10, gamma=0.01, kernel=sigmoid;, score=0.617 total time= 0.0s [CV 1/5] END .....C=10, gamma=0.001, kernel=rbf;, score=0.694 total time= 0.0s [CV 2/5] END .....C=10, gamma=0.001, kernel=rbf;, score=0.636 total time= 0.0s [CV 3/5] END .....C=10, gamma=0.001, kernel=rbf;, score=0.636 total time= 0.0s [CV 4/5] END .....C=10, gamma=0.001, kernel=rbf;, score=0.551 total time= 0.0s [CV 5/5] END .....C=10, gamma=0.001, kernel=rbf;, score=0.617 total time= 0.0s [CV 1/5] END .C=10, gamma=0.001, kernel=sigmoid;, score=0.685 total time= 0.0s [CV 2/5] END .C=10, gamma=0.001, kernel=sigmoid;, score=0.636 total time= 0.0s [CV 3/5] END .C=10, gamma=0.001, kernel=sigmoid;, score=0.617 total time= 0.0s [CV 4/5] END .C=10, gamma=0.001, kernel=sigmoid;, score=0.570 total time= 0.0s [CV 5/5] END .C=10, gamma=0.001, kernel=sigmoid;, score=0.579 total time= 0.0s [CV 1/5] END ........C=100, gamma=1, kernel=rbf;, score=0.546 total time= 0.0s [CV 2/5] END ........C=100, gamma=1, kernel=rbf;, score=0.598 total time= 0.0s [CV 3/5] END ........C=100, gamma=1, kernel=rbf;, score=0.495 total time= 0.0s [CV 4/5] END ........C=100, gamma=1, kernel=rbf;, score=0.523 total time= 0.0s [CV 5/5] END ........C=100, gamma=1, kernel=rbf;, score=0.570 total time= 0.0s [CV 1/5] END ....C=100, gamma=1, kernel=sigmoid;, score=0.593 total time= 0.0s [CV 2/5] END ....C=100, gamma=1, kernel=sigmoid;, score=0.533 total time= 0.0s [CV 3/5] END ....C=100, gamma=1, kernel=sigmoid;, score=0.607 total time= 0.0s [CV 4/5] END ....C=100, gamma=1, kernel=sigmoid;, score=0.570 total time= 0.0s [CV 5/5] END ....C=100, gamma=1, kernel=sigmoid;, score=0.495 total time= 0.0s [CV 1/5] END ......C=100, gamma=0.1, kernel=rbf;, score=0.556 total time= 0.0s [CV 2/5] END ......C=100, gamma=0.1, kernel=rbf;, score=0.570 total time= 0.0s [CV 3/5] END ......C=100, gamma=0.1, kernel=rbf;, score=0.561 total time= 0.0s [CV 4/5] END ......C=100, gamma=0.1, kernel=rbf;, score=0.551 total time= 0.0s [CV 5/5] END ......C=100, gamma=0.1, kernel=rbf;, score=0.477 total time= 0.0s [CV 1/5] END ..C=100, gamma=0.1, kernel=sigmoid;, score=0.556 total time= 0.0s [CV 2/5] END ..C=100, gamma=0.1, kernel=sigmoid;, score=0.570 total time= 0.0s [CV 3/5] END ..C=100, gamma=0.1, kernel=sigmoid;, score=0.551 total time= 0.0s [CV 4/5] END ..C=100, gamma=0.1, kernel=sigmoid;, score=0.589 total time= 0.0s [CV 5/5] END ..C=100, gamma=0.1, kernel=sigmoid;, score=0.486 total time= 0.0s [CV 1/5] END .....C=100, gamma=0.01, kernel=rbf;, score=0.676 total time= 0.0s [CV 2/5] END .....C=100, gamma=0.01, kernel=rbf;, score=0.589 total time= 0.0s [CV 3/5] END .....C=100, gamma=0.01, kernel=rbf;, score=0.617 total time= 0.0s [CV 4/5] END .....C=100, gamma=0.01, kernel=rbf;, score=0.514 total time= 0.0s [CV 5/5] END .....C=100, gamma=0.01, kernel=rbf;, score=0.579 total time= 0.0s [CV 1/5] END .C=100, gamma=0.01, kernel=sigmoid;, score=0.639 total time= 0.0s [CV 2/5] END .C=100, gamma=0.01, kernel=sigmoid;, score=0.607 total time= 0.0s [CV 3/5] END .C=100, gamma=0.01, kernel=sigmoid;, score=0.579 total time= 0.0s [CV 4/5] END .C=100, gamma=0.01, kernel=sigmoid;, score=0.579 total time= 0.0s [CV 5/5] END .C=100, gamma=0.01, kernel=sigmoid;, score=0.551 total time= 0.0s [CV 1/5] END ....C=100, gamma=0.001, kernel=rbf;, score=0.676 total time= 0.0s [CV 2/5] END ....C=100, gamma=0.001, kernel=rbf;, score=0.598 total time= 0.0s [CV 3/5] END ....C=100, gamma=0.001, kernel=rbf;, score=0.654 total time= 0.0s [CV 4/5] END ....C=100, gamma=0.001, kernel=rbf;, score=0.542 total time= 0.0s [CV 5/5] END ....C=100, gamma=0.001, kernel=rbf;, score=0.561 total time= 0.0s [CV 1/5] END C=100, gamma=0.001, kernel=sigmoid;, score=0.685 total time= 0.0s [CV 2/5] END C=100, gamma=0.001, kernel=sigmoid;, score=0.589 total time= 0.0s [CV 3/5] END C=100, gamma=0.001, kernel=sigmoid;, score=0.654 total time= 0.0s [CV 4/5] END C=100, gamma=0.001, kernel=sigmoid;, score=0.551 total time= 0.0s [CV 5/5] END C=100, gamma=0.001, kernel=sigmoid;, score=0.542 total time= 0.0s
GridSearchCV(estimator=SVC(class_weight='balanced'),
param_grid={'C': [0.1, 1, 10, 100], 'gamma': [1, 0.1, 0.01, 0.001],
'kernel': ['rbf', 'sigmoid']},
verbose=3)In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. GridSearchCV(estimator=SVC(class_weight='balanced'),
param_grid={'C': [0.1, 1, 10, 100], 'gamma': [1, 0.1, 0.01, 0.001],
'kernel': ['rbf', 'sigmoid']},
verbose=3)SVC(class_weight='balanced')
SVC(class_weight='balanced')
# training set
y_pred_svc_tr111 = SVC_grid.predict(X_train1)
accuracy_svc_tr111 = accuracy_score(y_train1, y_pred_svc_tr111)
print("Support Vector Classification Model Score (Training Set): %.2f%%" % (accuracy_svc_tr111 * 100.0))
# testing set
y_pred_svc_t111 = SVC_grid.predict(X_test1)
predictions_svc111 = [round(value) for value in y_pred_svc_t111]
# evaluate predictions
accuracy_svc_t111 = accuracy_score(y_test1, predictions_svc111)
print("Support Vector Classification Model Score (Testing Set): %.2f%%" % (accuracy_svc_t111 * 100.0))
Support Vector Classification Model Score (Training Set): 63.81% Support Vector Classification Model Score (Testing Set): 63.43%
plot_confusion_matrix(SVC_grid, X_test1, y_test1,cmap="YlGnBu");
# Printing the classification report
print(classification_report(y_test1, y_pred_svc_t111))
precision recall f1-score support
0 0.59 0.30 0.40 54
1 0.64 0.86 0.74 80
accuracy 0.63 134
macro avg 0.62 0.58 0.57 134
weighted avg 0.62 0.63 0.60 134
SVC_grid.fit(X_train2, y_train2)
Fitting 5 folds for each of 32 candidates, totalling 160 fits [CV 1/5] END ........C=0.1, gamma=1, kernel=rbf;, score=0.869 total time= 0.2s [CV 2/5] END ........C=0.1, gamma=1, kernel=rbf;, score=0.872 total time= 0.2s [CV 3/5] END ........C=0.1, gamma=1, kernel=rbf;, score=0.879 total time= 0.1s [CV 4/5] END ........C=0.1, gamma=1, kernel=rbf;, score=0.851 total time= 0.1s [CV 5/5] END ........C=0.1, gamma=1, kernel=rbf;, score=0.874 total time= 0.1s [CV 1/5] END ....C=0.1, gamma=1, kernel=sigmoid;, score=0.577 total time= 0.0s [CV 2/5] END ....C=0.1, gamma=1, kernel=sigmoid;, score=0.569 total time= 0.1s [CV 3/5] END ....C=0.1, gamma=1, kernel=sigmoid;, score=0.577 total time= 0.0s [CV 4/5] END ....C=0.1, gamma=1, kernel=sigmoid;, score=0.615 total time= 0.0s [CV 5/5] END ....C=0.1, gamma=1, kernel=sigmoid;, score=0.682 total time= 0.0s [CV 1/5] END ......C=0.1, gamma=0.1, kernel=rbf;, score=0.798 total time= 0.1s [CV 2/5] END ......C=0.1, gamma=0.1, kernel=rbf;, score=0.751 total time= 0.1s [CV 3/5] END ......C=0.1, gamma=0.1, kernel=rbf;, score=0.811 total time= 0.1s [CV 4/5] END ......C=0.1, gamma=0.1, kernel=rbf;, score=0.781 total time= 0.0s [CV 5/5] END ......C=0.1, gamma=0.1, kernel=rbf;, score=0.801 total time= 0.1s [CV 1/5] END ..C=0.1, gamma=0.1, kernel=sigmoid;, score=0.678 total time= 0.0s [CV 2/5] END ..C=0.1, gamma=0.1, kernel=sigmoid;, score=0.662 total time= 0.0s [CV 3/5] END ..C=0.1, gamma=0.1, kernel=sigmoid;, score=0.680 total time= 0.0s [CV 4/5] END ..C=0.1, gamma=0.1, kernel=sigmoid;, score=0.678 total time= 0.1s [CV 5/5] END ..C=0.1, gamma=0.1, kernel=sigmoid;, score=0.702 total time= 0.1s [CV 1/5] END .....C=0.1, gamma=0.01, kernel=rbf;, score=0.768 total time= 0.1s [CV 2/5] END .....C=0.1, gamma=0.01, kernel=rbf;, score=0.715 total time= 0.1s [CV 3/5] END .....C=0.1, gamma=0.01, kernel=rbf;, score=0.816 total time= 0.1s [CV 4/5] END .....C=0.1, gamma=0.01, kernel=rbf;, score=0.788 total time= 0.1s [CV 5/5] END .....C=0.1, gamma=0.01, kernel=rbf;, score=0.788 total time= 0.1s [CV 1/5] END .C=0.1, gamma=0.01, kernel=sigmoid;, score=0.768 total time= 0.1s [CV 2/5] END .C=0.1, gamma=0.01, kernel=sigmoid;, score=0.685 total time= 0.1s [CV 3/5] END .C=0.1, gamma=0.01, kernel=sigmoid;, score=0.831 total time= 0.1s [CV 4/5] END .C=0.1, gamma=0.01, kernel=sigmoid;, score=0.811 total time= 0.1s [CV 5/5] END .C=0.1, gamma=0.01, kernel=sigmoid;, score=0.816 total time= 0.1s [CV 1/5] END ....C=0.1, gamma=0.001, kernel=rbf;, score=0.469 total time= 0.1s [CV 2/5] END ....C=0.1, gamma=0.001, kernel=rbf;, score=0.411 total time= 0.1s [CV 3/5] END ....C=0.1, gamma=0.001, kernel=rbf;, score=0.519 total time= 0.1s [CV 4/5] END ....C=0.1, gamma=0.001, kernel=rbf;, score=0.496 total time= 0.1s [CV 5/5] END ....C=0.1, gamma=0.001, kernel=rbf;, score=0.503 total time= 0.1s [CV 1/5] END C=0.1, gamma=0.001, kernel=sigmoid;, score=0.285 total time= 0.1s [CV 2/5] END C=0.1, gamma=0.001, kernel=sigmoid;, score=0.305 total time= 0.1s [CV 3/5] END C=0.1, gamma=0.001, kernel=sigmoid;, score=0.370 total time= 0.1s [CV 4/5] END C=0.1, gamma=0.001, kernel=sigmoid;, score=0.345 total time= 0.1s [CV 5/5] END C=0.1, gamma=0.001, kernel=sigmoid;, score=0.348 total time= 0.1s [CV 1/5] END ..........C=1, gamma=1, kernel=rbf;, score=0.854 total time= 0.0s [CV 2/5] END ..........C=1, gamma=1, kernel=rbf;, score=0.856 total time= 0.0s [CV 3/5] END ..........C=1, gamma=1, kernel=rbf;, score=0.866 total time= 0.0s [CV 4/5] END ..........C=1, gamma=1, kernel=rbf;, score=0.869 total time= 0.0s [CV 5/5] END ..........C=1, gamma=1, kernel=rbf;, score=0.859 total time= 0.0s [CV 1/5] END ......C=1, gamma=1, kernel=sigmoid;, score=0.564 total time= 0.0s [CV 2/5] END ......C=1, gamma=1, kernel=sigmoid;, score=0.562 total time= 0.0s [CV 3/5] END ......C=1, gamma=1, kernel=sigmoid;, score=0.557 total time= 0.0s [CV 4/5] END ......C=1, gamma=1, kernel=sigmoid;, score=0.615 total time= 0.0s [CV 5/5] END ......C=1, gamma=1, kernel=sigmoid;, score=0.631 total time= 0.0s [CV 1/5] END ........C=1, gamma=0.1, kernel=rbf;, score=0.816 total time= 0.0s [CV 2/5] END ........C=1, gamma=0.1, kernel=rbf;, score=0.783 total time= 0.0s [CV 3/5] END ........C=1, gamma=0.1, kernel=rbf;, score=0.831 total time= 0.0s [CV 4/5] END ........C=1, gamma=0.1, kernel=rbf;, score=0.766 total time= 0.0s [CV 5/5] END ........C=1, gamma=0.1, kernel=rbf;, score=0.773 total time= 0.0s [CV 1/5] END ....C=1, gamma=0.1, kernel=sigmoid;, score=0.622 total time= 0.0s [CV 2/5] END ....C=1, gamma=0.1, kernel=sigmoid;, score=0.675 total time= 0.0s [CV 3/5] END ....C=1, gamma=0.1, kernel=sigmoid;, score=0.615 total time= 0.0s [CV 4/5] END ....C=1, gamma=0.1, kernel=sigmoid;, score=0.594 total time= 0.0s [CV 5/5] END ....C=1, gamma=0.1, kernel=sigmoid;, score=0.596 total time= 0.0s [CV 1/5] END .......C=1, gamma=0.01, kernel=rbf;, score=0.736 total time= 0.1s [CV 2/5] END .......C=1, gamma=0.01, kernel=rbf;, score=0.693 total time= 0.0s [CV 3/5] END .......C=1, gamma=0.01, kernel=rbf;, score=0.776 total time= 0.1s [CV 4/5] END .......C=1, gamma=0.01, kernel=rbf;, score=0.751 total time= 0.0s [CV 5/5] END .......C=1, gamma=0.01, kernel=rbf;, score=0.725 total time= 0.0s [CV 1/5] END ...C=1, gamma=0.01, kernel=sigmoid;, score=0.751 total time= 0.0s [CV 2/5] END ...C=1, gamma=0.01, kernel=sigmoid;, score=0.720 total time= 0.0s [CV 3/5] END ...C=1, gamma=0.01, kernel=sigmoid;, score=0.766 total time= 0.0s [CV 4/5] END ...C=1, gamma=0.01, kernel=sigmoid;, score=0.710 total time= 0.0s [CV 5/5] END ...C=1, gamma=0.01, kernel=sigmoid;, score=0.740 total time= 0.0s [CV 1/5] END ......C=1, gamma=0.001, kernel=rbf;, score=0.771 total time= 0.1s [CV 2/5] END ......C=1, gamma=0.001, kernel=rbf;, score=0.733 total time= 0.1s [CV 3/5] END ......C=1, gamma=0.001, kernel=rbf;, score=0.826 total time= 0.1s [CV 4/5] END ......C=1, gamma=0.001, kernel=rbf;, score=0.804 total time= 0.1s [CV 5/5] END ......C=1, gamma=0.001, kernel=rbf;, score=0.806 total time= 0.1s [CV 1/5] END ..C=1, gamma=0.001, kernel=sigmoid;, score=0.768 total time= 0.0s [CV 2/5] END ..C=1, gamma=0.001, kernel=sigmoid;, score=0.685 total time= 0.0s [CV 3/5] END ..C=1, gamma=0.001, kernel=sigmoid;, score=0.831 total time= 0.0s [CV 4/5] END ..C=1, gamma=0.001, kernel=sigmoid;, score=0.811 total time= 0.0s [CV 5/5] END ..C=1, gamma=0.001, kernel=sigmoid;, score=0.818 total time= 0.0s [CV 1/5] END .........C=10, gamma=1, kernel=rbf;, score=0.892 total time= 0.0s [CV 2/5] END .........C=10, gamma=1, kernel=rbf;, score=0.892 total time= 0.0s [CV 3/5] END .........C=10, gamma=1, kernel=rbf;, score=0.907 total time= 0.0s [CV 4/5] END .........C=10, gamma=1, kernel=rbf;, score=0.894 total time= 0.0s [CV 5/5] END .........C=10, gamma=1, kernel=rbf;, score=0.907 total time= 0.0s [CV 1/5] END .....C=10, gamma=1, kernel=sigmoid;, score=0.562 total time= 0.0s [CV 2/5] END .....C=10, gamma=1, kernel=sigmoid;, score=0.562 total time= 0.0s [CV 3/5] END .....C=10, gamma=1, kernel=sigmoid;, score=0.557 total time= 0.0s [CV 4/5] END .....C=10, gamma=1, kernel=sigmoid;, score=0.615 total time= 0.0s [CV 5/5] END .....C=10, gamma=1, kernel=sigmoid;, score=0.636 total time= 0.0s [CV 1/5] END .......C=10, gamma=0.1, kernel=rbf;, score=0.829 total time= 0.0s [CV 2/5] END .......C=10, gamma=0.1, kernel=rbf;, score=0.816 total time= 0.0s [CV 3/5] END .......C=10, gamma=0.1, kernel=rbf;, score=0.829 total time= 0.0s [CV 4/5] END .......C=10, gamma=0.1, kernel=rbf;, score=0.834 total time= 0.0s [CV 5/5] END .......C=10, gamma=0.1, kernel=rbf;, score=0.803 total time= 0.0s [CV 1/5] END ...C=10, gamma=0.1, kernel=sigmoid;, score=0.620 total time= 0.0s [CV 2/5] END ...C=10, gamma=0.1, kernel=sigmoid;, score=0.660 total time= 0.0s [CV 3/5] END ...C=10, gamma=0.1, kernel=sigmoid;, score=0.615 total time= 0.0s [CV 4/5] END ...C=10, gamma=0.1, kernel=sigmoid;, score=0.592 total time= 0.0s [CV 5/5] END ...C=10, gamma=0.1, kernel=sigmoid;, score=0.586 total time= 0.0s [CV 1/5] END ......C=10, gamma=0.01, kernel=rbf;, score=0.788 total time= 0.0s [CV 2/5] END ......C=10, gamma=0.01, kernel=rbf;, score=0.743 total time= 0.0s [CV 3/5] END ......C=10, gamma=0.01, kernel=rbf;, score=0.771 total time= 0.0s [CV 4/5] END ......C=10, gamma=0.01, kernel=rbf;, score=0.728 total time= 0.1s [CV 5/5] END ......C=10, gamma=0.01, kernel=rbf;, score=0.730 total time= 0.0s [CV 1/5] END ..C=10, gamma=0.01, kernel=sigmoid;, score=0.743 total time= 0.0s [CV 2/5] END ..C=10, gamma=0.01, kernel=sigmoid;, score=0.723 total time= 0.0s [CV 3/5] END ..C=10, gamma=0.01, kernel=sigmoid;, score=0.738 total time= 0.0s [CV 4/5] END ..C=10, gamma=0.01, kernel=sigmoid;, score=0.688 total time= 0.0s [CV 5/5] END ..C=10, gamma=0.01, kernel=sigmoid;, score=0.692 total time= 0.0s [CV 1/5] END .....C=10, gamma=0.001, kernel=rbf;, score=0.748 total time= 0.0s [CV 2/5] END .....C=10, gamma=0.001, kernel=rbf;, score=0.678 total time= 0.0s [CV 3/5] END .....C=10, gamma=0.001, kernel=rbf;, score=0.738 total time= 0.0s [CV 4/5] END .....C=10, gamma=0.001, kernel=rbf;, score=0.698 total time= 0.0s [CV 5/5] END .....C=10, gamma=0.001, kernel=rbf;, score=0.694 total time= 0.1s [CV 1/5] END .C=10, gamma=0.001, kernel=sigmoid;, score=0.743 total time= 0.0s [CV 2/5] END .C=10, gamma=0.001, kernel=sigmoid;, score=0.700 total time= 0.0s [CV 3/5] END .C=10, gamma=0.001, kernel=sigmoid;, score=0.766 total time= 0.0s [CV 4/5] END .C=10, gamma=0.001, kernel=sigmoid;, score=0.718 total time= 0.0s [CV 5/5] END .C=10, gamma=0.001, kernel=sigmoid;, score=0.727 total time= 0.0s [CV 1/5] END ........C=100, gamma=1, kernel=rbf;, score=0.907 total time= 0.0s [CV 2/5] END ........C=100, gamma=1, kernel=rbf;, score=0.907 total time= 0.0s [CV 3/5] END ........C=100, gamma=1, kernel=rbf;, score=0.919 total time= 0.0s [CV 4/5] END ........C=100, gamma=1, kernel=rbf;, score=0.907 total time= 0.0s [CV 5/5] END ........C=100, gamma=1, kernel=rbf;, score=0.912 total time= 0.0s [CV 1/5] END ....C=100, gamma=1, kernel=sigmoid;, score=0.562 total time= 0.0s [CV 2/5] END ....C=100, gamma=1, kernel=sigmoid;, score=0.564 total time= 0.0s [CV 3/5] END ....C=100, gamma=1, kernel=sigmoid;, score=0.557 total time= 0.0s [CV 4/5] END ....C=100, gamma=1, kernel=sigmoid;, score=0.612 total time= 0.0s [CV 5/5] END ....C=100, gamma=1, kernel=sigmoid;, score=0.636 total time= 0.0s [CV 1/5] END ......C=100, gamma=0.1, kernel=rbf;, score=0.841 total time= 0.0s [CV 2/5] END ......C=100, gamma=0.1, kernel=rbf;, score=0.841 total time= 0.0s [CV 3/5] END ......C=100, gamma=0.1, kernel=rbf;, score=0.834 total time= 0.0s [CV 4/5] END ......C=100, gamma=0.1, kernel=rbf;, score=0.841 total time= 0.0s [CV 5/5] END ......C=100, gamma=0.1, kernel=rbf;, score=0.836 total time= 0.0s [CV 1/5] END ..C=100, gamma=0.1, kernel=sigmoid;, score=0.622 total time= 0.0s [CV 2/5] END ..C=100, gamma=0.1, kernel=sigmoid;, score=0.657 total time= 0.0s [CV 3/5] END ..C=100, gamma=0.1, kernel=sigmoid;, score=0.625 total time= 0.0s [CV 4/5] END ..C=100, gamma=0.1, kernel=sigmoid;, score=0.584 total time= 0.0s [CV 5/5] END ..C=100, gamma=0.1, kernel=sigmoid;, score=0.591 total time= 0.0s [CV 1/5] END .....C=100, gamma=0.01, kernel=rbf;, score=0.806 total time= 0.1s [CV 2/5] END .....C=100, gamma=0.01, kernel=rbf;, score=0.758 total time= 0.1s [CV 3/5] END .....C=100, gamma=0.01, kernel=rbf;, score=0.806 total time= 0.1s [CV 4/5] END .....C=100, gamma=0.01, kernel=rbf;, score=0.746 total time= 0.1s [CV 5/5] END .....C=100, gamma=0.01, kernel=rbf;, score=0.747 total time= 0.1s [CV 1/5] END .C=100, gamma=0.01, kernel=sigmoid;, score=0.685 total time= 0.0s [CV 2/5] END .C=100, gamma=0.01, kernel=sigmoid;, score=0.683 total time= 0.0s [CV 3/5] END .C=100, gamma=0.01, kernel=sigmoid;, score=0.685 total time= 0.0s [CV 4/5] END .C=100, gamma=0.01, kernel=sigmoid;, score=0.647 total time= 0.0s [CV 5/5] END .C=100, gamma=0.01, kernel=sigmoid;, score=0.636 total time= 0.0s [CV 1/5] END ....C=100, gamma=0.001, kernel=rbf;, score=0.741 total time= 0.1s [CV 2/5] END ....C=100, gamma=0.001, kernel=rbf;, score=0.703 total time= 0.1s [CV 3/5] END ....C=100, gamma=0.001, kernel=rbf;, score=0.738 total time= 0.1s [CV 4/5] END ....C=100, gamma=0.001, kernel=rbf;, score=0.710 total time= 0.1s [CV 5/5] END ....C=100, gamma=0.001, kernel=rbf;, score=0.707 total time= 0.1s [CV 1/5] END C=100, gamma=0.001, kernel=sigmoid;, score=0.736 total time= 0.0s [CV 2/5] END C=100, gamma=0.001, kernel=sigmoid;, score=0.685 total time= 0.0s [CV 3/5] END C=100, gamma=0.001, kernel=sigmoid;, score=0.743 total time= 0.0s [CV 4/5] END C=100, gamma=0.001, kernel=sigmoid;, score=0.683 total time= 0.0s [CV 5/5] END C=100, gamma=0.001, kernel=sigmoid;, score=0.689 total time= 0.0s
GridSearchCV(estimator=SVC(class_weight='balanced'),
param_grid={'C': [0.1, 1, 10, 100], 'gamma': [1, 0.1, 0.01, 0.001],
'kernel': ['rbf', 'sigmoid']},
verbose=3)In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. GridSearchCV(estimator=SVC(class_weight='balanced'),
param_grid={'C': [0.1, 1, 10, 100], 'gamma': [1, 0.1, 0.01, 0.001],
'kernel': ['rbf', 'sigmoid']},
verbose=3)SVC(class_weight='balanced')
SVC(class_weight='balanced')
# training set
y_pred_svc_tr222 = SVC_grid.predict(X_train2)
accuracy_svc_tr222 = accuracy_score(y_train2, y_pred_svc_tr222)
print("Support Vector Classification Model Score (Training Set): %.2f%%" % (accuracy_svc_tr222 * 100.0))
# testing set
y_pred_svc_t222 = SVC_grid.predict(X_test2)
predictions_svc222 = [round(value) for value in y_pred_svc_t222]
# evaluate predictions
accuracy_svc_t222 = accuracy_score(y_test2, predictions_svc222)
print("Support Vector Classification Model Score (Testing Set): %.2f%%" % (accuracy_svc_t222 * 100.0))
Support Vector Classification Model Score (Training Set): 99.55% Support Vector Classification Model Score (Testing Set): 90.34%
plot_confusion_matrix(SVC_grid, X_test2, y_test2,cmap="YlGnBu");
# Printing the classification report
print(classification_report(y_test2, y_pred_svc_t222))
precision recall f1-score support
0 0.95 0.94 0.95 468
1 0.23 0.28 0.25 29
accuracy 0.90 497
macro avg 0.59 0.61 0.60 497
weighted avg 0.91 0.90 0.91 497
SVC_grid.fit(X_train3, y_train3)
Fitting 5 folds for each of 32 candidates, totalling 160 fits [CV 1/5] END ........C=0.1, gamma=1, kernel=rbf;, score=0.889 total time= 0.2s [CV 2/5] END ........C=0.1, gamma=1, kernel=rbf;, score=0.864 total time= 0.3s [CV 3/5] END ........C=0.1, gamma=1, kernel=rbf;, score=0.904 total time= 0.2s [CV 4/5] END ........C=0.1, gamma=1, kernel=rbf;, score=0.854 total time= 0.3s [CV 5/5] END ........C=0.1, gamma=1, kernel=rbf;, score=0.854 total time= 0.2s [CV 1/5] END ....C=0.1, gamma=1, kernel=sigmoid;, score=0.531 total time= 0.1s [CV 2/5] END ....C=0.1, gamma=1, kernel=sigmoid;, score=0.479 total time= 0.1s [CV 3/5] END ....C=0.1, gamma=1, kernel=sigmoid;, score=0.481 total time= 0.1s [CV 4/5] END ....C=0.1, gamma=1, kernel=sigmoid;, score=0.448 total time= 0.2s [CV 5/5] END ....C=0.1, gamma=1, kernel=sigmoid;, score=0.491 total time= 0.1s [CV 1/5] END ......C=0.1, gamma=0.1, kernel=rbf;, score=0.678 total time= 0.1s [CV 2/5] END ......C=0.1, gamma=0.1, kernel=rbf;, score=0.657 total time= 0.1s [CV 3/5] END ......C=0.1, gamma=0.1, kernel=rbf;, score=0.657 total time= 0.1s [CV 4/5] END ......C=0.1, gamma=0.1, kernel=rbf;, score=0.680 total time= 0.1s [CV 5/5] END ......C=0.1, gamma=0.1, kernel=rbf;, score=0.652 total time= 0.1s [CV 1/5] END ..C=0.1, gamma=0.1, kernel=sigmoid;, score=0.484 total time= 0.0s [CV 2/5] END ..C=0.1, gamma=0.1, kernel=sigmoid;, score=0.484 total time= 0.1s [CV 3/5] END ..C=0.1, gamma=0.1, kernel=sigmoid;, score=0.461 total time= 0.0s [CV 4/5] END ..C=0.1, gamma=0.1, kernel=sigmoid;, score=0.481 total time= 0.0s [CV 5/5] END ..C=0.1, gamma=0.1, kernel=sigmoid;, score=0.534 total time= 0.0s [CV 1/5] END .....C=0.1, gamma=0.01, kernel=rbf;, score=0.673 total time= 0.1s [CV 2/5] END .....C=0.1, gamma=0.01, kernel=rbf;, score=0.549 total time= 0.1s [CV 3/5] END .....C=0.1, gamma=0.01, kernel=rbf;, score=0.647 total time= 0.1s [CV 4/5] END .....C=0.1, gamma=0.01, kernel=rbf;, score=0.393 total time= 0.1s [CV 5/5] END .....C=0.1, gamma=0.01, kernel=rbf;, score=0.557 total time= 0.1s [CV 1/5] END .C=0.1, gamma=0.01, kernel=sigmoid;, score=0.695 total time= 0.1s [CV 2/5] END .C=0.1, gamma=0.01, kernel=sigmoid;, score=0.161 total time= 0.1s [CV 3/5] END .C=0.1, gamma=0.01, kernel=sigmoid;, score=0.411 total time= 0.1s [CV 4/5] END .C=0.1, gamma=0.01, kernel=sigmoid;, score=0.350 total time= 0.1s [CV 5/5] END .C=0.1, gamma=0.01, kernel=sigmoid;, score=0.212 total time= 0.1s [CV 1/5] END ....C=0.1, gamma=0.001, kernel=rbf;, score=0.962 total time= 0.1s [CV 2/5] END ....C=0.1, gamma=0.001, kernel=rbf;, score=0.030 total time= 0.1s [CV 3/5] END ....C=0.1, gamma=0.001, kernel=rbf;, score=0.030 total time= 0.1s [CV 4/5] END ....C=0.1, gamma=0.001, kernel=rbf;, score=0.030 total time= 0.1s [CV 5/5] END ....C=0.1, gamma=0.001, kernel=rbf;, score=0.030 total time= 0.1s [CV 1/5] END C=0.1, gamma=0.001, kernel=sigmoid;, score=0.967 total time= 0.1s [CV 2/5] END C=0.1, gamma=0.001, kernel=sigmoid;, score=0.030 total time= 0.1s [CV 3/5] END C=0.1, gamma=0.001, kernel=sigmoid;, score=0.030 total time= 0.1s [CV 4/5] END C=0.1, gamma=0.001, kernel=sigmoid;, score=0.030 total time= 0.1s [CV 5/5] END C=0.1, gamma=0.001, kernel=sigmoid;, score=0.030 total time= 0.1s [CV 1/5] END ..........C=1, gamma=1, kernel=rbf;, score=0.929 total time= 0.0s [CV 2/5] END ..........C=1, gamma=1, kernel=rbf;, score=0.929 total time= 0.0s [CV 3/5] END ..........C=1, gamma=1, kernel=rbf;, score=0.942 total time= 0.0s [CV 4/5] END ..........C=1, gamma=1, kernel=rbf;, score=0.922 total time= 0.0s [CV 5/5] END ..........C=1, gamma=1, kernel=rbf;, score=0.935 total time= 0.0s [CV 1/5] END ......C=1, gamma=1, kernel=sigmoid;, score=0.537 total time= 0.1s [CV 2/5] END ......C=1, gamma=1, kernel=sigmoid;, score=0.461 total time= 0.1s [CV 3/5] END ......C=1, gamma=1, kernel=sigmoid;, score=0.486 total time= 0.0s [CV 4/5] END ......C=1, gamma=1, kernel=sigmoid;, score=0.436 total time= 0.1s [CV 5/5] END ......C=1, gamma=1, kernel=sigmoid;, score=0.463 total time= 0.1s [CV 1/5] END ........C=1, gamma=0.1, kernel=rbf;, score=0.733 total time= 0.0s [CV 2/5] END ........C=1, gamma=0.1, kernel=rbf;, score=0.673 total time= 0.0s [CV 3/5] END ........C=1, gamma=0.1, kernel=rbf;, score=0.708 total time= 0.1s [CV 4/5] END ........C=1, gamma=0.1, kernel=rbf;, score=0.708 total time= 0.0s [CV 5/5] END ........C=1, gamma=0.1, kernel=rbf;, score=0.710 total time= 0.1s [CV 1/5] END ....C=1, gamma=0.1, kernel=sigmoid;, score=0.516 total time= 0.0s [CV 2/5] END ....C=1, gamma=0.1, kernel=sigmoid;, score=0.494 total time= 0.0s [CV 3/5] END ....C=1, gamma=0.1, kernel=sigmoid;, score=0.484 total time= 0.0s [CV 4/5] END ....C=1, gamma=0.1, kernel=sigmoid;, score=0.484 total time= 0.0s [CV 5/5] END ....C=1, gamma=0.1, kernel=sigmoid;, score=0.587 total time= 0.0s [CV 1/5] END .......C=1, gamma=0.01, kernel=rbf;, score=0.635 total time= 0.1s [CV 2/5] END .......C=1, gamma=0.01, kernel=rbf;, score=0.572 total time= 0.1s [CV 3/5] END .......C=1, gamma=0.01, kernel=rbf;, score=0.612 total time= 0.1s [CV 4/5] END .......C=1, gamma=0.01, kernel=rbf;, score=0.602 total time= 0.1s [CV 5/5] END .......C=1, gamma=0.01, kernel=rbf;, score=0.572 total time= 0.1s [CV 1/5] END ...C=1, gamma=0.01, kernel=sigmoid;, score=0.534 total time= 0.0s [CV 2/5] END ...C=1, gamma=0.01, kernel=sigmoid;, score=0.509 total time= 0.1s [CV 3/5] END ...C=1, gamma=0.01, kernel=sigmoid;, score=0.594 total time= 0.0s [CV 4/5] END ...C=1, gamma=0.01, kernel=sigmoid;, score=0.542 total time= 0.1s [CV 5/5] END ...C=1, gamma=0.01, kernel=sigmoid;, score=0.469 total time= 0.0s [CV 1/5] END ......C=1, gamma=0.001, kernel=rbf;, score=0.683 total time= 0.1s [CV 2/5] END ......C=1, gamma=0.001, kernel=rbf;, score=0.466 total time= 0.1s [CV 3/5] END ......C=1, gamma=0.001, kernel=rbf;, score=0.577 total time= 0.1s [CV 4/5] END ......C=1, gamma=0.001, kernel=rbf;, score=0.461 total time= 0.1s [CV 5/5] END ......C=1, gamma=0.001, kernel=rbf;, score=0.463 total time= 0.1s [CV 1/5] END ..C=1, gamma=0.001, kernel=sigmoid;, score=0.695 total time= 0.1s [CV 2/5] END ..C=1, gamma=0.001, kernel=sigmoid;, score=0.164 total time= 0.1s [CV 3/5] END ..C=1, gamma=0.001, kernel=sigmoid;, score=0.413 total time= 0.1s [CV 4/5] END ..C=1, gamma=0.001, kernel=sigmoid;, score=0.338 total time= 0.1s [CV 5/5] END ..C=1, gamma=0.001, kernel=sigmoid;, score=0.199 total time= 0.1s [CV 1/5] END .........C=10, gamma=1, kernel=rbf;, score=0.942 total time= 0.0s [CV 2/5] END .........C=10, gamma=1, kernel=rbf;, score=0.947 total time= 0.0s [CV 3/5] END .........C=10, gamma=1, kernel=rbf;, score=0.950 total time= 0.0s [CV 4/5] END .........C=10, gamma=1, kernel=rbf;, score=0.945 total time= 0.0s [CV 5/5] END .........C=10, gamma=1, kernel=rbf;, score=0.957 total time= 0.0s [CV 1/5] END .....C=10, gamma=1, kernel=sigmoid;, score=0.531 total time= 0.1s [CV 2/5] END .....C=10, gamma=1, kernel=sigmoid;, score=0.453 total time= 0.1s [CV 3/5] END .....C=10, gamma=1, kernel=sigmoid;, score=0.486 total time= 0.1s [CV 4/5] END .....C=10, gamma=1, kernel=sigmoid;, score=0.431 total time= 0.1s [CV 5/5] END .....C=10, gamma=1, kernel=sigmoid;, score=0.463 total time= 0.1s [CV 1/5] END .......C=10, gamma=0.1, kernel=rbf;, score=0.831 total time= 0.0s [CV 2/5] END .......C=10, gamma=0.1, kernel=rbf;, score=0.816 total time= 0.1s [CV 3/5] END .......C=10, gamma=0.1, kernel=rbf;, score=0.791 total time= 0.0s [CV 4/5] END .......C=10, gamma=0.1, kernel=rbf;, score=0.771 total time= 0.1s [CV 5/5] END .......C=10, gamma=0.1, kernel=rbf;, score=0.793 total time= 0.0s [CV 1/5] END ...C=10, gamma=0.1, kernel=sigmoid;, score=0.501 total time= 0.0s [CV 2/5] END ...C=10, gamma=0.1, kernel=sigmoid;, score=0.549 total time= 0.0s [CV 3/5] END ...C=10, gamma=0.1, kernel=sigmoid;, score=0.499 total time= 0.0s [CV 4/5] END ...C=10, gamma=0.1, kernel=sigmoid;, score=0.489 total time= 0.1s [CV 5/5] END ...C=10, gamma=0.1, kernel=sigmoid;, score=0.582 total time= 0.0s [CV 1/5] END ......C=10, gamma=0.01, kernel=rbf;, score=0.670 total time= 0.1s [CV 2/5] END ......C=10, gamma=0.01, kernel=rbf;, score=0.592 total time= 0.1s [CV 3/5] END ......C=10, gamma=0.01, kernel=rbf;, score=0.630 total time= 0.1s [CV 4/5] END ......C=10, gamma=0.01, kernel=rbf;, score=0.645 total time= 0.1s [CV 5/5] END ......C=10, gamma=0.01, kernel=rbf;, score=0.605 total time= 0.1s [CV 1/5] END ..C=10, gamma=0.01, kernel=sigmoid;, score=0.481 total time= 0.1s [CV 2/5] END ..C=10, gamma=0.01, kernel=sigmoid;, score=0.559 total time= 0.1s [CV 3/5] END ..C=10, gamma=0.01, kernel=sigmoid;, score=0.549 total time= 0.1s [CV 4/5] END ..C=10, gamma=0.01, kernel=sigmoid;, score=0.496 total time= 0.1s [CV 5/5] END ..C=10, gamma=0.01, kernel=sigmoid;, score=0.463 total time= 0.1s [CV 1/5] END .....C=10, gamma=0.001, kernel=rbf;, score=0.610 total time= 0.1s [CV 2/5] END .....C=10, gamma=0.001, kernel=rbf;, score=0.509 total time= 0.1s [CV 3/5] END .....C=10, gamma=0.001, kernel=rbf;, score=0.564 total time= 0.1s [CV 4/5] END .....C=10, gamma=0.001, kernel=rbf;, score=0.539 total time= 0.1s [CV 5/5] END .....C=10, gamma=0.001, kernel=rbf;, score=0.509 total time= 0.1s [CV 1/5] END .C=10, gamma=0.001, kernel=sigmoid;, score=0.547 total time= 0.0s [CV 2/5] END .C=10, gamma=0.001, kernel=sigmoid;, score=0.484 total time= 0.1s [CV 3/5] END .C=10, gamma=0.001, kernel=sigmoid;, score=0.592 total time= 0.0s [CV 4/5] END .C=10, gamma=0.001, kernel=sigmoid;, score=0.537 total time= 0.1s [CV 5/5] END .C=10, gamma=0.001, kernel=sigmoid;, score=0.476 total time= 0.0s [CV 1/5] END ........C=100, gamma=1, kernel=rbf;, score=0.947 total time= 0.0s [CV 2/5] END ........C=100, gamma=1, kernel=rbf;, score=0.952 total time= 0.0s [CV 3/5] END ........C=100, gamma=1, kernel=rbf;, score=0.950 total time= 0.0s [CV 4/5] END ........C=100, gamma=1, kernel=rbf;, score=0.945 total time= 0.0s [CV 5/5] END ........C=100, gamma=1, kernel=rbf;, score=0.952 total time= 0.0s [CV 1/5] END ....C=100, gamma=1, kernel=sigmoid;, score=0.531 total time= 0.1s [CV 2/5] END ....C=100, gamma=1, kernel=sigmoid;, score=0.453 total time= 0.1s [CV 3/5] END ....C=100, gamma=1, kernel=sigmoid;, score=0.486 total time= 0.1s [CV 4/5] END ....C=100, gamma=1, kernel=sigmoid;, score=0.431 total time= 0.1s [CV 5/5] END ....C=100, gamma=1, kernel=sigmoid;, score=0.463 total time= 0.1s [CV 1/5] END ......C=100, gamma=0.1, kernel=rbf;, score=0.882 total time= 0.1s [CV 2/5] END ......C=100, gamma=0.1, kernel=rbf;, score=0.897 total time= 0.0s [CV 3/5] END ......C=100, gamma=0.1, kernel=rbf;, score=0.894 total time= 0.0s [CV 4/5] END ......C=100, gamma=0.1, kernel=rbf;, score=0.872 total time= 0.0s [CV 5/5] END ......C=100, gamma=0.1, kernel=rbf;, score=0.892 total time= 0.0s [CV 1/5] END ..C=100, gamma=0.1, kernel=sigmoid;, score=0.499 total time= 0.0s [CV 2/5] END ..C=100, gamma=0.1, kernel=sigmoid;, score=0.544 total time= 0.0s [CV 3/5] END ..C=100, gamma=0.1, kernel=sigmoid;, score=0.504 total time= 0.0s [CV 4/5] END ..C=100, gamma=0.1, kernel=sigmoid;, score=0.486 total time= 0.0s [CV 5/5] END ..C=100, gamma=0.1, kernel=sigmoid;, score=0.584 total time= 0.0s [CV 1/5] END .....C=100, gamma=0.01, kernel=rbf;, score=0.660 total time= 0.1s [CV 2/5] END .....C=100, gamma=0.01, kernel=rbf;, score=0.587 total time= 0.1s [CV 3/5] END .....C=100, gamma=0.01, kernel=rbf;, score=0.660 total time= 0.1s [CV 4/5] END .....C=100, gamma=0.01, kernel=rbf;, score=0.640 total time= 0.1s [CV 5/5] END .....C=100, gamma=0.01, kernel=rbf;, score=0.705 total time= 0.1s [CV 1/5] END .C=100, gamma=0.01, kernel=sigmoid;, score=0.592 total time= 0.0s [CV 2/5] END .C=100, gamma=0.01, kernel=sigmoid;, score=0.577 total time= 0.0s [CV 3/5] END .C=100, gamma=0.01, kernel=sigmoid;, score=0.554 total time= 0.0s [CV 4/5] END .C=100, gamma=0.01, kernel=sigmoid;, score=0.537 total time= 0.0s [CV 5/5] END .C=100, gamma=0.01, kernel=sigmoid;, score=0.599 total time= 0.0s [CV 1/5] END ....C=100, gamma=0.001, kernel=rbf;, score=0.612 total time= 0.1s [CV 2/5] END ....C=100, gamma=0.001, kernel=rbf;, score=0.552 total time= 0.1s [CV 3/5] END ....C=100, gamma=0.001, kernel=rbf;, score=0.625 total time= 0.1s [CV 4/5] END ....C=100, gamma=0.001, kernel=rbf;, score=0.607 total time= 0.1s [CV 5/5] END ....C=100, gamma=0.001, kernel=rbf;, score=0.567 total time= 0.1s [CV 1/5] END C=100, gamma=0.001, kernel=sigmoid;, score=0.607 total time= 0.1s [CV 2/5] END C=100, gamma=0.001, kernel=sigmoid;, score=0.524 total time= 0.1s [CV 3/5] END C=100, gamma=0.001, kernel=sigmoid;, score=0.562 total time= 0.1s [CV 4/5] END C=100, gamma=0.001, kernel=sigmoid;, score=0.534 total time= 0.1s [CV 5/5] END C=100, gamma=0.001, kernel=sigmoid;, score=0.499 total time= 0.1s
GridSearchCV(estimator=SVC(class_weight='balanced'),
param_grid={'C': [0.1, 1, 10, 100], 'gamma': [1, 0.1, 0.01, 0.001],
'kernel': ['rbf', 'sigmoid']},
verbose=3)In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. GridSearchCV(estimator=SVC(class_weight='balanced'),
param_grid={'C': [0.1, 1, 10, 100], 'gamma': [1, 0.1, 0.01, 0.001],
'kernel': ['rbf', 'sigmoid']},
verbose=3)SVC(class_weight='balanced')
SVC(class_weight='balanced')
# training set
y_pred_svc_tr333 = SVC_grid.predict(X_train3)
accuracy_svc_tr333 = accuracy_score(y_train3, y_pred_svc_tr333)
print("Support Vector Classification Model Score (Training Set): %.2f%%" % (accuracy_svc_tr333 * 100.0))
# testing set
y_pred_svc_t333 = SVC_grid.predict(X_test3)
predictions_svc333 = [round(value) for value in y_pred_svc_t333]
# evaluate predictions
accuracy_svc_t333 = accuracy_score(y_test3, predictions_svc333)
print("Support Vector Classification Model Score (Testing Set): %.2f%%" % (accuracy_svc_t333 * 100.0))
Support Vector Classification Model Score (Training Set): 100.00% Support Vector Classification Model Score (Testing Set): 93.36%
plot_confusion_matrix(SVC_grid, X_test3, y_test3,cmap="YlGnBu");
# Printing the classification report
print(classification_report(y_test3, y_pred_svc_t333))
precision recall f1-score support
0 0.00 0.00 0.00 17
1 0.96 0.97 0.97 480
accuracy 0.93 497
macro avg 0.48 0.48 0.48 497
weighted avg 0.93 0.93 0.93 497
SVC_grid.fit(X_train4, y_train4)
Fitting 5 folds for each of 32 candidates, totalling 160 fits [CV 1/5] END ........C=0.1, gamma=1, kernel=rbf;, score=0.645 total time= 0.2s [CV 2/5] END ........C=0.1, gamma=1, kernel=rbf;, score=0.668 total time= 0.1s [CV 3/5] END ........C=0.1, gamma=1, kernel=rbf;, score=0.675 total time= 0.1s [CV 4/5] END ........C=0.1, gamma=1, kernel=rbf;, score=0.632 total time= 0.2s [CV 5/5] END ........C=0.1, gamma=1, kernel=rbf;, score=0.657 total time= 0.1s [CV 1/5] END ....C=0.1, gamma=1, kernel=sigmoid;, score=0.501 total time= 0.1s [CV 2/5] END ....C=0.1, gamma=1, kernel=sigmoid;, score=0.504 total time= 0.1s [CV 3/5] END ....C=0.1, gamma=1, kernel=sigmoid;, score=0.526 total time= 0.1s [CV 4/5] END ....C=0.1, gamma=1, kernel=sigmoid;, score=0.514 total time= 0.1s [CV 5/5] END ....C=0.1, gamma=1, kernel=sigmoid;, score=0.523 total time= 0.1s [CV 1/5] END ......C=0.1, gamma=0.1, kernel=rbf;, score=0.670 total time= 0.1s [CV 2/5] END ......C=0.1, gamma=0.1, kernel=rbf;, score=0.703 total time= 0.1s [CV 3/5] END ......C=0.1, gamma=0.1, kernel=rbf;, score=0.668 total time= 0.2s [CV 4/5] END ......C=0.1, gamma=0.1, kernel=rbf;, score=0.642 total time= 0.2s [CV 5/5] END ......C=0.1, gamma=0.1, kernel=rbf;, score=0.674 total time= 0.2s [CV 1/5] END ..C=0.1, gamma=0.1, kernel=sigmoid;, score=0.559 total time= 0.1s [CV 2/5] END ..C=0.1, gamma=0.1, kernel=sigmoid;, score=0.645 total time= 0.1s [CV 3/5] END ..C=0.1, gamma=0.1, kernel=sigmoid;, score=0.635 total time= 0.1s [CV 4/5] END ..C=0.1, gamma=0.1, kernel=sigmoid;, score=0.594 total time= 0.1s [CV 5/5] END ..C=0.1, gamma=0.1, kernel=sigmoid;, score=0.561 total time= 0.1s [CV 1/5] END .....C=0.1, gamma=0.01, kernel=rbf;, score=0.650 total time= 0.2s [CV 2/5] END .....C=0.1, gamma=0.01, kernel=rbf;, score=0.670 total time= 0.2s [CV 3/5] END .....C=0.1, gamma=0.01, kernel=rbf;, score=0.645 total time= 0.2s [CV 4/5] END .....C=0.1, gamma=0.01, kernel=rbf;, score=0.612 total time= 0.2s [CV 5/5] END .....C=0.1, gamma=0.01, kernel=rbf;, score=0.641 total time= 0.1s [CV 1/5] END .C=0.1, gamma=0.01, kernel=sigmoid;, score=0.622 total time= 0.1s [CV 2/5] END .C=0.1, gamma=0.01, kernel=sigmoid;, score=0.615 total time= 0.1s [CV 3/5] END .C=0.1, gamma=0.01, kernel=sigmoid;, score=0.625 total time= 0.1s [CV 4/5] END .C=0.1, gamma=0.01, kernel=sigmoid;, score=0.589 total time= 0.1s [CV 5/5] END .C=0.1, gamma=0.01, kernel=sigmoid;, score=0.662 total time= 0.1s [CV 1/5] END ....C=0.1, gamma=0.001, kernel=rbf;, score=0.547 total time= 0.2s [CV 2/5] END ....C=0.1, gamma=0.001, kernel=rbf;, score=0.539 total time= 0.2s [CV 3/5] END ....C=0.1, gamma=0.001, kernel=rbf;, score=0.559 total time= 0.2s [CV 4/5] END ....C=0.1, gamma=0.001, kernel=rbf;, score=0.562 total time= 0.1s [CV 5/5] END ....C=0.1, gamma=0.001, kernel=rbf;, score=0.609 total time= 0.1s [CV 1/5] END C=0.1, gamma=0.001, kernel=sigmoid;, score=0.433 total time= 0.1s [CV 2/5] END C=0.1, gamma=0.001, kernel=sigmoid;, score=0.443 total time= 0.1s [CV 3/5] END C=0.1, gamma=0.001, kernel=sigmoid;, score=0.448 total time= 0.1s [CV 4/5] END C=0.1, gamma=0.001, kernel=sigmoid;, score=0.436 total time= 0.1s [CV 5/5] END C=0.1, gamma=0.001, kernel=sigmoid;, score=0.581 total time= 0.1s [CV 1/5] END ..........C=1, gamma=1, kernel=rbf;, score=0.660 total time= 0.1s [CV 2/5] END ..........C=1, gamma=1, kernel=rbf;, score=0.700 total time= 0.1s [CV 3/5] END ..........C=1, gamma=1, kernel=rbf;, score=0.665 total time= 0.1s [CV 4/5] END ..........C=1, gamma=1, kernel=rbf;, score=0.612 total time= 0.1s [CV 5/5] END ..........C=1, gamma=1, kernel=rbf;, score=0.649 total time= 0.1s [CV 1/5] END ......C=1, gamma=1, kernel=sigmoid;, score=0.516 total time= 0.0s [CV 2/5] END ......C=1, gamma=1, kernel=sigmoid;, score=0.509 total time= 0.1s [CV 3/5] END ......C=1, gamma=1, kernel=sigmoid;, score=0.521 total time= 0.1s [CV 4/5] END ......C=1, gamma=1, kernel=sigmoid;, score=0.509 total time= 0.1s [CV 5/5] END ......C=1, gamma=1, kernel=sigmoid;, score=0.523 total time= 0.0s [CV 1/5] END ........C=1, gamma=0.1, kernel=rbf;, score=0.693 total time= 0.1s [CV 2/5] END ........C=1, gamma=0.1, kernel=rbf;, score=0.708 total time= 0.1s [CV 3/5] END ........C=1, gamma=0.1, kernel=rbf;, score=0.680 total time= 0.1s [CV 4/5] END ........C=1, gamma=0.1, kernel=rbf;, score=0.645 total time= 0.1s [CV 5/5] END ........C=1, gamma=0.1, kernel=rbf;, score=0.687 total time= 0.1s [CV 1/5] END ....C=1, gamma=0.1, kernel=sigmoid;, score=0.544 total time= 0.0s [CV 2/5] END ....C=1, gamma=0.1, kernel=sigmoid;, score=0.602 total time= 0.0s [CV 3/5] END ....C=1, gamma=0.1, kernel=sigmoid;, score=0.582 total time= 0.0s [CV 4/5] END ....C=1, gamma=0.1, kernel=sigmoid;, score=0.562 total time= 0.0s [CV 5/5] END ....C=1, gamma=0.1, kernel=sigmoid;, score=0.535 total time= 0.0s [CV 1/5] END .......C=1, gamma=0.01, kernel=rbf;, score=0.698 total time= 0.1s [CV 2/5] END .......C=1, gamma=0.01, kernel=rbf;, score=0.698 total time= 0.1s [CV 3/5] END .......C=1, gamma=0.01, kernel=rbf;, score=0.662 total time= 0.1s [CV 4/5] END .......C=1, gamma=0.01, kernel=rbf;, score=0.647 total time= 0.1s [CV 5/5] END .......C=1, gamma=0.01, kernel=rbf;, score=0.679 total time= 0.1s [CV 1/5] END ...C=1, gamma=0.01, kernel=sigmoid;, score=0.630 total time= 0.1s [CV 2/5] END ...C=1, gamma=0.01, kernel=sigmoid;, score=0.673 total time= 0.0s [CV 3/5] END ...C=1, gamma=0.01, kernel=sigmoid;, score=0.647 total time= 0.1s [CV 4/5] END ...C=1, gamma=0.01, kernel=sigmoid;, score=0.612 total time= 0.0s [CV 5/5] END ...C=1, gamma=0.01, kernel=sigmoid;, score=0.662 total time= 0.1s [CV 1/5] END ......C=1, gamma=0.001, kernel=rbf;, score=0.637 total time= 0.1s [CV 2/5] END ......C=1, gamma=0.001, kernel=rbf;, score=0.665 total time= 0.1s [CV 3/5] END ......C=1, gamma=0.001, kernel=rbf;, score=0.652 total time= 0.1s [CV 4/5] END ......C=1, gamma=0.001, kernel=rbf;, score=0.605 total time= 0.1s [CV 5/5] END ......C=1, gamma=0.001, kernel=rbf;, score=0.662 total time= 0.1s [CV 1/5] END ..C=1, gamma=0.001, kernel=sigmoid;, score=0.622 total time= 0.1s [CV 2/5] END ..C=1, gamma=0.001, kernel=sigmoid;, score=0.620 total time= 0.1s [CV 3/5] END ..C=1, gamma=0.001, kernel=sigmoid;, score=0.627 total time= 0.1s [CV 4/5] END ..C=1, gamma=0.001, kernel=sigmoid;, score=0.594 total time= 0.1s [CV 5/5] END ..C=1, gamma=0.001, kernel=sigmoid;, score=0.662 total time= 0.1s [CV 1/5] END .........C=10, gamma=1, kernel=rbf;, score=0.657 total time= 0.1s [CV 2/5] END .........C=10, gamma=1, kernel=rbf;, score=0.625 total time= 0.1s [CV 3/5] END .........C=10, gamma=1, kernel=rbf;, score=0.610 total time= 0.1s [CV 4/5] END .........C=10, gamma=1, kernel=rbf;, score=0.597 total time= 0.1s [CV 5/5] END .........C=10, gamma=1, kernel=rbf;, score=0.604 total time= 0.1s [CV 1/5] END .....C=10, gamma=1, kernel=sigmoid;, score=0.496 total time= 0.1s [CV 2/5] END .....C=10, gamma=1, kernel=sigmoid;, score=0.509 total time= 0.1s [CV 3/5] END .....C=10, gamma=1, kernel=sigmoid;, score=0.524 total time= 0.0s [CV 4/5] END .....C=10, gamma=1, kernel=sigmoid;, score=0.509 total time= 0.0s [CV 5/5] END .....C=10, gamma=1, kernel=sigmoid;, score=0.525 total time= 0.0s [CV 1/5] END .......C=10, gamma=0.1, kernel=rbf;, score=0.673 total time= 0.1s [CV 2/5] END .......C=10, gamma=0.1, kernel=rbf;, score=0.710 total time= 0.1s [CV 3/5] END .......C=10, gamma=0.1, kernel=rbf;, score=0.673 total time= 0.1s [CV 4/5] END .......C=10, gamma=0.1, kernel=rbf;, score=0.625 total time= 0.1s [CV 5/5] END .......C=10, gamma=0.1, kernel=rbf;, score=0.662 total time= 0.1s [CV 1/5] END ...C=10, gamma=0.1, kernel=sigmoid;, score=0.537 total time= 0.0s [CV 2/5] END ...C=10, gamma=0.1, kernel=sigmoid;, score=0.602 total time= 0.0s [CV 3/5] END ...C=10, gamma=0.1, kernel=sigmoid;, score=0.579 total time= 0.0s [CV 4/5] END ...C=10, gamma=0.1, kernel=sigmoid;, score=0.564 total time= 0.0s [CV 5/5] END ...C=10, gamma=0.1, kernel=sigmoid;, score=0.538 total time= 0.0s [CV 1/5] END ......C=10, gamma=0.01, kernel=rbf;, score=0.695 total time= 0.1s [CV 2/5] END ......C=10, gamma=0.01, kernel=rbf;, score=0.710 total time= 0.1s [CV 3/5] END ......C=10, gamma=0.01, kernel=rbf;, score=0.693 total time= 0.1s [CV 4/5] END ......C=10, gamma=0.01, kernel=rbf;, score=0.632 total time= 0.1s [CV 5/5] END ......C=10, gamma=0.01, kernel=rbf;, score=0.684 total time= 0.2s [CV 1/5] END ..C=10, gamma=0.01, kernel=sigmoid;, score=0.673 total time= 0.1s [CV 2/5] END ..C=10, gamma=0.01, kernel=sigmoid;, score=0.650 total time= 0.2s [CV 3/5] END ..C=10, gamma=0.01, kernel=sigmoid;, score=0.620 total time= 0.1s [CV 4/5] END ..C=10, gamma=0.01, kernel=sigmoid;, score=0.587 total time= 0.1s [CV 5/5] END ..C=10, gamma=0.01, kernel=sigmoid;, score=0.639 total time= 0.2s [CV 1/5] END .....C=10, gamma=0.001, kernel=rbf;, score=0.647 total time= 0.1s [CV 2/5] END .....C=10, gamma=0.001, kernel=rbf;, score=0.685 total time= 0.1s [CV 3/5] END .....C=10, gamma=0.001, kernel=rbf;, score=0.650 total time= 0.1s [CV 4/5] END .....C=10, gamma=0.001, kernel=rbf;, score=0.615 total time= 0.1s [CV 5/5] END .....C=10, gamma=0.001, kernel=rbf;, score=0.667 total time= 0.2s [CV 1/5] END .C=10, gamma=0.001, kernel=sigmoid;, score=0.632 total time= 0.0s [CV 2/5] END .C=10, gamma=0.001, kernel=sigmoid;, score=0.675 total time= 0.1s [CV 3/5] END .C=10, gamma=0.001, kernel=sigmoid;, score=0.642 total time= 0.1s [CV 4/5] END .C=10, gamma=0.001, kernel=sigmoid;, score=0.605 total time= 0.1s [CV 5/5] END .C=10, gamma=0.001, kernel=sigmoid;, score=0.659 total time= 0.0s [CV 1/5] END ........C=100, gamma=1, kernel=rbf;, score=0.617 total time= 0.2s [CV 2/5] END ........C=100, gamma=1, kernel=rbf;, score=0.594 total time= 0.2s [CV 3/5] END ........C=100, gamma=1, kernel=rbf;, score=0.579 total time= 0.2s [CV 4/5] END ........C=100, gamma=1, kernel=rbf;, score=0.622 total time= 0.2s [CV 5/5] END ........C=100, gamma=1, kernel=rbf;, score=0.588 total time= 0.2s [CV 1/5] END ....C=100, gamma=1, kernel=sigmoid;, score=0.496 total time= 0.0s [CV 2/5] END ....C=100, gamma=1, kernel=sigmoid;, score=0.509 total time= 0.0s [CV 3/5] END ....C=100, gamma=1, kernel=sigmoid;, score=0.524 total time= 0.0s [CV 4/5] END ....C=100, gamma=1, kernel=sigmoid;, score=0.509 total time= 0.0s [CV 5/5] END ....C=100, gamma=1, kernel=sigmoid;, score=0.525 total time= 0.0s [CV 1/5] END ......C=100, gamma=0.1, kernel=rbf;, score=0.637 total time= 0.2s [CV 2/5] END ......C=100, gamma=0.1, kernel=rbf;, score=0.683 total time= 0.2s [CV 3/5] END ......C=100, gamma=0.1, kernel=rbf;, score=0.657 total time= 0.2s [CV 4/5] END ......C=100, gamma=0.1, kernel=rbf;, score=0.605 total time= 0.3s [CV 5/5] END ......C=100, gamma=0.1, kernel=rbf;, score=0.636 total time= 0.2s [CV 1/5] END ..C=100, gamma=0.1, kernel=sigmoid;, score=0.537 total time= 0.0s [CV 2/5] END ..C=100, gamma=0.1, kernel=sigmoid;, score=0.602 total time= 0.0s [CV 3/5] END ..C=100, gamma=0.1, kernel=sigmoid;, score=0.572 total time= 0.0s [CV 4/5] END ..C=100, gamma=0.1, kernel=sigmoid;, score=0.552 total time= 0.0s [CV 5/5] END ..C=100, gamma=0.1, kernel=sigmoid;, score=0.535 total time= 0.0s [CV 1/5] END .....C=100, gamma=0.01, kernel=rbf;, score=0.683 total time= 0.1s [CV 2/5] END .....C=100, gamma=0.01, kernel=rbf;, score=0.715 total time= 0.1s [CV 3/5] END .....C=100, gamma=0.01, kernel=rbf;, score=0.688 total time= 0.1s [CV 4/5] END .....C=100, gamma=0.01, kernel=rbf;, score=0.650 total time= 0.1s [CV 5/5] END .....C=100, gamma=0.01, kernel=rbf;, score=0.697 total time= 0.1s [CV 1/5] END .C=100, gamma=0.01, kernel=sigmoid;, score=0.630 total time= 0.0s [CV 2/5] END .C=100, gamma=0.01, kernel=sigmoid;, score=0.610 total time= 0.0s [CV 3/5] END .C=100, gamma=0.01, kernel=sigmoid;, score=0.607 total time= 0.0s [CV 4/5] END .C=100, gamma=0.01, kernel=sigmoid;, score=0.587 total time= 0.0s [CV 5/5] END .C=100, gamma=0.01, kernel=sigmoid;, score=0.598 total time= 0.0s [CV 1/5] END ....C=100, gamma=0.001, kernel=rbf;, score=0.695 total time= 0.1s [CV 2/5] END ....C=100, gamma=0.001, kernel=rbf;, score=0.710 total time= 0.1s [CV 3/5] END ....C=100, gamma=0.001, kernel=rbf;, score=0.668 total time= 0.1s [CV 4/5] END ....C=100, gamma=0.001, kernel=rbf;, score=0.642 total time= 0.1s [CV 5/5] END ....C=100, gamma=0.001, kernel=rbf;, score=0.677 total time= 0.1s [CV 1/5] END C=100, gamma=0.001, kernel=sigmoid;, score=0.647 total time= 0.1s [CV 2/5] END C=100, gamma=0.001, kernel=sigmoid;, score=0.678 total time= 0.1s [CV 3/5] END C=100, gamma=0.001, kernel=sigmoid;, score=0.647 total time= 0.1s [CV 4/5] END C=100, gamma=0.001, kernel=sigmoid;, score=0.607 total time= 0.1s [CV 5/5] END C=100, gamma=0.001, kernel=sigmoid;, score=0.669 total time= 0.1s
GridSearchCV(estimator=SVC(class_weight='balanced'),
param_grid={'C': [0.1, 1, 10, 100], 'gamma': [1, 0.1, 0.01, 0.001],
'kernel': ['rbf', 'sigmoid']},
verbose=3)In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. GridSearchCV(estimator=SVC(class_weight='balanced'),
param_grid={'C': [0.1, 1, 10, 100], 'gamma': [1, 0.1, 0.01, 0.001],
'kernel': ['rbf', 'sigmoid']},
verbose=3)SVC(class_weight='balanced')
SVC(class_weight='balanced')
# training set
y_pred_svc_tr444 = SVC_grid.predict(X_train4)
accuracy_svc_tr444 = accuracy_score(y_train4, y_pred_svc_tr444)
print("Support Vector Classification Model Score (Training Set): %.2f%%" % (accuracy_svc_tr44 * 100.0))
# testing set
y_pred_svc_t444 = SVC_grid.predict(X_test4)
predictions_svc444 = [round(value) for value in y_pred_svc_t444]
# evaluate predictions
accuracy_svc_t444 = accuracy_score(y_test4, predictions_svc444)
print("Support Vector Classification Model Score (Testing Set): %.2f%%" % (accuracy_svc_t444 * 100.0))
Support Vector Classification Model Score (Training Set): 65.78% Support Vector Classification Model Score (Testing Set): 66.00%
plot_confusion_matrix(SVC_grid, X_test4, y_test4,cmap="YlGnBu");
# Printing the classification report
print(classification_report(y_test4, y_pred_svc_t444))
precision recall f1-score support
0 0.62 0.63 0.62 222
1 0.70 0.68 0.69 275
accuracy 0.66 497
macro avg 0.66 0.66 0.66 497
weighted avg 0.66 0.66 0.66 497
SVC_grid.fit(X_train5, y_train5)
Fitting 5 folds for each of 32 candidates, totalling 160 fits [CV 1/5] END ........C=0.1, gamma=1, kernel=rbf;, score=0.683 total time= 0.2s [CV 2/5] END ........C=0.1, gamma=1, kernel=rbf;, score=0.690 total time= 0.1s [CV 3/5] END ........C=0.1, gamma=1, kernel=rbf;, score=0.708 total time= 0.1s [CV 4/5] END ........C=0.1, gamma=1, kernel=rbf;, score=0.728 total time= 0.1s [CV 5/5] END ........C=0.1, gamma=1, kernel=rbf;, score=0.678 total time= 0.1s [CV 1/5] END ....C=0.1, gamma=1, kernel=sigmoid;, score=0.516 total time= 0.1s [CV 2/5] END ....C=0.1, gamma=1, kernel=sigmoid;, score=0.491 total time= 0.1s [CV 3/5] END ....C=0.1, gamma=1, kernel=sigmoid;, score=0.489 total time= 0.1s [CV 4/5] END ....C=0.1, gamma=1, kernel=sigmoid;, score=0.491 total time= 0.1s [CV 5/5] END ....C=0.1, gamma=1, kernel=sigmoid;, score=0.451 total time= 0.1s [CV 1/5] END ......C=0.1, gamma=0.1, kernel=rbf;, score=0.673 total time= 0.1s [CV 2/5] END ......C=0.1, gamma=0.1, kernel=rbf;, score=0.675 total time= 0.1s [CV 3/5] END ......C=0.1, gamma=0.1, kernel=rbf;, score=0.657 total time= 0.1s [CV 4/5] END ......C=0.1, gamma=0.1, kernel=rbf;, score=0.695 total time= 0.1s [CV 5/5] END ......C=0.1, gamma=0.1, kernel=rbf;, score=0.650 total time= 0.1s [CV 1/5] END ..C=0.1, gamma=0.1, kernel=sigmoid;, score=0.562 total time= 0.0s [CV 2/5] END ..C=0.1, gamma=0.1, kernel=sigmoid;, score=0.599 total time= 0.0s [CV 3/5] END ..C=0.1, gamma=0.1, kernel=sigmoid;, score=0.564 total time= 0.0s [CV 4/5] END ..C=0.1, gamma=0.1, kernel=sigmoid;, score=0.559 total time= 0.1s [CV 5/5] END ..C=0.1, gamma=0.1, kernel=sigmoid;, score=0.557 total time= 0.1s [CV 1/5] END .....C=0.1, gamma=0.01, kernel=rbf;, score=0.630 total time= 0.1s [CV 2/5] END .....C=0.1, gamma=0.01, kernel=rbf;, score=0.637 total time= 0.1s [CV 3/5] END .....C=0.1, gamma=0.01, kernel=rbf;, score=0.627 total time= 0.1s [CV 4/5] END .....C=0.1, gamma=0.01, kernel=rbf;, score=0.650 total time= 0.1s [CV 5/5] END .....C=0.1, gamma=0.01, kernel=rbf;, score=0.645 total time= 0.1s [CV 1/5] END .C=0.1, gamma=0.01, kernel=sigmoid;, score=0.620 total time= 0.1s [CV 2/5] END .C=0.1, gamma=0.01, kernel=sigmoid;, score=0.632 total time= 0.1s [CV 3/5] END .C=0.1, gamma=0.01, kernel=sigmoid;, score=0.630 total time= 0.0s [CV 4/5] END .C=0.1, gamma=0.01, kernel=sigmoid;, score=0.642 total time= 0.1s [CV 5/5] END .C=0.1, gamma=0.01, kernel=sigmoid;, score=0.645 total time= 0.0s [CV 1/5] END ....C=0.1, gamma=0.001, kernel=rbf;, score=0.494 total time= 0.1s [CV 2/5] END ....C=0.1, gamma=0.001, kernel=rbf;, score=0.484 total time= 0.1s [CV 3/5] END ....C=0.1, gamma=0.001, kernel=rbf;, score=0.484 total time= 0.1s [CV 4/5] END ....C=0.1, gamma=0.001, kernel=rbf;, score=0.481 total time= 0.1s [CV 5/5] END ....C=0.1, gamma=0.001, kernel=rbf;, score=0.537 total time= 0.1s [CV 1/5] END C=0.1, gamma=0.001, kernel=sigmoid;, score=0.630 total time= 0.1s [CV 2/5] END C=0.1, gamma=0.001, kernel=sigmoid;, score=0.393 total time= 0.1s [CV 3/5] END C=0.1, gamma=0.001, kernel=sigmoid;, score=0.398 total time= 0.1s [CV 4/5] END C=0.1, gamma=0.001, kernel=sigmoid;, score=0.385 total time= 0.1s [CV 5/5] END C=0.1, gamma=0.001, kernel=sigmoid;, score=0.375 total time= 0.1s [CV 1/5] END ..........C=1, gamma=1, kernel=rbf;, score=0.683 total time= 0.1s [CV 2/5] END ..........C=1, gamma=1, kernel=rbf;, score=0.698 total time= 0.1s [CV 3/5] END ..........C=1, gamma=1, kernel=rbf;, score=0.698 total time= 0.1s [CV 4/5] END ..........C=1, gamma=1, kernel=rbf;, score=0.715 total time= 0.1s [CV 5/5] END ..........C=1, gamma=1, kernel=rbf;, score=0.683 total time= 0.1s [CV 1/5] END ......C=1, gamma=1, kernel=sigmoid;, score=0.514 total time= 0.0s [CV 2/5] END ......C=1, gamma=1, kernel=sigmoid;, score=0.499 total time= 0.0s [CV 3/5] END ......C=1, gamma=1, kernel=sigmoid;, score=0.474 total time= 0.0s [CV 4/5] END ......C=1, gamma=1, kernel=sigmoid;, score=0.491 total time= 0.0s [CV 5/5] END ......C=1, gamma=1, kernel=sigmoid;, score=0.443 total time= 0.1s [CV 1/5] END ........C=1, gamma=0.1, kernel=rbf;, score=0.705 total time= 0.1s [CV 2/5] END ........C=1, gamma=0.1, kernel=rbf;, score=0.670 total time= 0.1s [CV 3/5] END ........C=1, gamma=0.1, kernel=rbf;, score=0.695 total time= 0.1s [CV 4/5] END ........C=1, gamma=0.1, kernel=rbf;, score=0.693 total time= 0.1s [CV 5/5] END ........C=1, gamma=0.1, kernel=rbf;, score=0.678 total time= 0.1s [CV 1/5] END ....C=1, gamma=0.1, kernel=sigmoid;, score=0.542 total time= 0.0s [CV 2/5] END ....C=1, gamma=0.1, kernel=sigmoid;, score=0.572 total time= 0.0s [CV 3/5] END ....C=1, gamma=0.1, kernel=sigmoid;, score=0.542 total time= 0.0s [CV 4/5] END ....C=1, gamma=0.1, kernel=sigmoid;, score=0.544 total time= 0.0s [CV 5/5] END ....C=1, gamma=0.1, kernel=sigmoid;, score=0.539 total time= 0.0s [CV 1/5] END .......C=1, gamma=0.01, kernel=rbf;, score=0.650 total time= 0.1s [CV 2/5] END .......C=1, gamma=0.01, kernel=rbf;, score=0.647 total time= 0.1s [CV 3/5] END .......C=1, gamma=0.01, kernel=rbf;, score=0.650 total time= 0.2s [CV 4/5] END .......C=1, gamma=0.01, kernel=rbf;, score=0.657 total time= 0.2s [CV 5/5] END .......C=1, gamma=0.01, kernel=rbf;, score=0.660 total time= 0.1s [CV 1/5] END ...C=1, gamma=0.01, kernel=sigmoid;, score=0.655 total time= 0.1s [CV 2/5] END ...C=1, gamma=0.01, kernel=sigmoid;, score=0.640 total time= 0.1s [CV 3/5] END ...C=1, gamma=0.01, kernel=sigmoid;, score=0.630 total time= 0.1s [CV 4/5] END ...C=1, gamma=0.01, kernel=sigmoid;, score=0.655 total time= 0.1s [CV 5/5] END ...C=1, gamma=0.01, kernel=sigmoid;, score=0.632 total time= 0.1s [CV 1/5] END ......C=1, gamma=0.001, kernel=rbf;, score=0.632 total time= 0.2s [CV 2/5] END ......C=1, gamma=0.001, kernel=rbf;, score=0.637 total time= 0.2s [CV 3/5] END ......C=1, gamma=0.001, kernel=rbf;, score=0.622 total time= 0.1s [CV 4/5] END ......C=1, gamma=0.001, kernel=rbf;, score=0.642 total time= 0.1s [CV 5/5] END ......C=1, gamma=0.001, kernel=rbf;, score=0.632 total time= 0.2s [CV 1/5] END ..C=1, gamma=0.001, kernel=sigmoid;, score=0.620 total time= 0.1s [CV 2/5] END ..C=1, gamma=0.001, kernel=sigmoid;, score=0.632 total time= 0.1s [CV 3/5] END ..C=1, gamma=0.001, kernel=sigmoid;, score=0.627 total time= 0.0s [CV 4/5] END ..C=1, gamma=0.001, kernel=sigmoid;, score=0.640 total time= 0.1s [CV 5/5] END ..C=1, gamma=0.001, kernel=sigmoid;, score=0.645 total time= 0.2s [CV 1/5] END .........C=10, gamma=1, kernel=rbf;, score=0.632 total time= 0.2s [CV 2/5] END .........C=10, gamma=1, kernel=rbf;, score=0.650 total time= 0.2s [CV 3/5] END .........C=10, gamma=1, kernel=rbf;, score=0.700 total time= 0.2s [CV 4/5] END .........C=10, gamma=1, kernel=rbf;, score=0.675 total time= 0.2s [CV 5/5] END .........C=10, gamma=1, kernel=rbf;, score=0.668 total time= 0.2s [CV 1/5] END .....C=10, gamma=1, kernel=sigmoid;, score=0.511 total time= 0.1s [CV 2/5] END .....C=10, gamma=1, kernel=sigmoid;, score=0.504 total time= 0.1s [CV 3/5] END .....C=10, gamma=1, kernel=sigmoid;, score=0.474 total time= 0.1s [CV 4/5] END .....C=10, gamma=1, kernel=sigmoid;, score=0.494 total time= 0.1s [CV 5/5] END .....C=10, gamma=1, kernel=sigmoid;, score=0.443 total time= 0.1s [CV 1/5] END .......C=10, gamma=0.1, kernel=rbf;, score=0.678 total time= 0.1s [CV 2/5] END .......C=10, gamma=0.1, kernel=rbf;, score=0.678 total time= 0.1s [CV 3/5] END .......C=10, gamma=0.1, kernel=rbf;, score=0.710 total time= 0.1s [CV 4/5] END .......C=10, gamma=0.1, kernel=rbf;, score=0.698 total time= 0.1s [CV 5/5] END .......C=10, gamma=0.1, kernel=rbf;, score=0.675 total time= 0.1s [CV 1/5] END ...C=10, gamma=0.1, kernel=sigmoid;, score=0.542 total time= 0.0s [CV 2/5] END ...C=10, gamma=0.1, kernel=sigmoid;, score=0.567 total time= 0.0s [CV 3/5] END ...C=10, gamma=0.1, kernel=sigmoid;, score=0.544 total time= 0.0s [CV 4/5] END ...C=10, gamma=0.1, kernel=sigmoid;, score=0.539 total time= 0.0s [CV 5/5] END ...C=10, gamma=0.1, kernel=sigmoid;, score=0.539 total time= 0.0s [CV 1/5] END ......C=10, gamma=0.01, kernel=rbf;, score=0.688 total time= 0.1s [CV 2/5] END ......C=10, gamma=0.01, kernel=rbf;, score=0.678 total time= 0.1s [CV 3/5] END ......C=10, gamma=0.01, kernel=rbf;, score=0.675 total time= 0.1s [CV 4/5] END ......C=10, gamma=0.01, kernel=rbf;, score=0.668 total time= 0.2s [CV 5/5] END ......C=10, gamma=0.01, kernel=rbf;, score=0.668 total time= 0.1s [CV 1/5] END ..C=10, gamma=0.01, kernel=sigmoid;, score=0.642 total time= 0.0s [CV 2/5] END ..C=10, gamma=0.01, kernel=sigmoid;, score=0.655 total time= 0.1s [CV 3/5] END ..C=10, gamma=0.01, kernel=sigmoid;, score=0.652 total time= 0.2s [CV 4/5] END ..C=10, gamma=0.01, kernel=sigmoid;, score=0.627 total time= 0.0s [CV 5/5] END ..C=10, gamma=0.01, kernel=sigmoid;, score=0.635 total time= 0.0s [CV 1/5] END .....C=10, gamma=0.001, kernel=rbf;, score=0.645 total time= 0.1s [CV 2/5] END .....C=10, gamma=0.001, kernel=rbf;, score=0.647 total time= 0.1s [CV 3/5] END .....C=10, gamma=0.001, kernel=rbf;, score=0.627 total time= 0.1s [CV 4/5] END .....C=10, gamma=0.001, kernel=rbf;, score=0.652 total time= 0.2s [CV 5/5] END .....C=10, gamma=0.001, kernel=rbf;, score=0.635 total time= 0.4s [CV 1/5] END .C=10, gamma=0.001, kernel=sigmoid;, score=0.645 total time= 0.1s [CV 2/5] END .C=10, gamma=0.001, kernel=sigmoid;, score=0.642 total time= 0.1s [CV 3/5] END .C=10, gamma=0.001, kernel=sigmoid;, score=0.630 total time= 0.0s [CV 4/5] END .C=10, gamma=0.001, kernel=sigmoid;, score=0.647 total time= 0.1s [CV 5/5] END .C=10, gamma=0.001, kernel=sigmoid;, score=0.632 total time= 0.1s [CV 1/5] END ........C=100, gamma=1, kernel=rbf;, score=0.617 total time= 0.2s [CV 2/5] END ........C=100, gamma=1, kernel=rbf;, score=0.635 total time= 0.2s [CV 3/5] END ........C=100, gamma=1, kernel=rbf;, score=0.678 total time= 0.2s [CV 4/5] END ........C=100, gamma=1, kernel=rbf;, score=0.662 total time= 0.2s [CV 5/5] END ........C=100, gamma=1, kernel=rbf;, score=0.657 total time= 0.2s [CV 1/5] END ....C=100, gamma=1, kernel=sigmoid;, score=0.514 total time= 0.0s [CV 2/5] END ....C=100, gamma=1, kernel=sigmoid;, score=0.504 total time= 0.0s [CV 3/5] END ....C=100, gamma=1, kernel=sigmoid;, score=0.474 total time= 0.1s [CV 4/5] END ....C=100, gamma=1, kernel=sigmoid;, score=0.494 total time= 0.0s [CV 5/5] END ....C=100, gamma=1, kernel=sigmoid;, score=0.443 total time= 0.1s [CV 1/5] END ......C=100, gamma=0.1, kernel=rbf;, score=0.645 total time= 0.2s [CV 2/5] END ......C=100, gamma=0.1, kernel=rbf;, score=0.675 total time= 0.2s [CV 3/5] END ......C=100, gamma=0.1, kernel=rbf;, score=0.685 total time= 0.2s [CV 4/5] END ......C=100, gamma=0.1, kernel=rbf;, score=0.713 total time= 0.2s [CV 5/5] END ......C=100, gamma=0.1, kernel=rbf;, score=0.652 total time= 0.2s [CV 1/5] END ..C=100, gamma=0.1, kernel=sigmoid;, score=0.544 total time= 0.0s [CV 2/5] END ..C=100, gamma=0.1, kernel=sigmoid;, score=0.567 total time= 0.0s [CV 3/5] END ..C=100, gamma=0.1, kernel=sigmoid;, score=0.542 total time= 0.0s [CV 4/5] END ..C=100, gamma=0.1, kernel=sigmoid;, score=0.542 total time= 0.0s [CV 5/5] END ..C=100, gamma=0.1, kernel=sigmoid;, score=0.539 total time= 0.0s [CV 1/5] END .....C=100, gamma=0.01, kernel=rbf;, score=0.705 total time= 0.2s [CV 2/5] END .....C=100, gamma=0.01, kernel=rbf;, score=0.660 total time= 0.1s [CV 3/5] END .....C=100, gamma=0.01, kernel=rbf;, score=0.673 total time= 0.2s [CV 4/5] END .....C=100, gamma=0.01, kernel=rbf;, score=0.688 total time= 0.2s [CV 5/5] END .....C=100, gamma=0.01, kernel=rbf;, score=0.675 total time= 0.1s [CV 1/5] END .C=100, gamma=0.01, kernel=sigmoid;, score=0.620 total time= 0.0s [CV 2/5] END .C=100, gamma=0.01, kernel=sigmoid;, score=0.635 total time= 0.0s [CV 3/5] END .C=100, gamma=0.01, kernel=sigmoid;, score=0.617 total time= 0.0s [CV 4/5] END .C=100, gamma=0.01, kernel=sigmoid;, score=0.615 total time= 0.0s [CV 5/5] END .C=100, gamma=0.01, kernel=sigmoid;, score=0.599 total time= 0.0s [CV 1/5] END ....C=100, gamma=0.001, kernel=rbf;, score=0.652 total time= 0.2s [CV 2/5] END ....C=100, gamma=0.001, kernel=rbf;, score=0.657 total time= 0.1s [CV 3/5] END ....C=100, gamma=0.001, kernel=rbf;, score=0.657 total time= 0.1s [CV 4/5] END ....C=100, gamma=0.001, kernel=rbf;, score=0.652 total time= 0.1s [CV 5/5] END ....C=100, gamma=0.001, kernel=rbf;, score=0.657 total time= 0.1s [CV 1/5] END C=100, gamma=0.001, kernel=sigmoid;, score=0.647 total time= 0.0s [CV 2/5] END C=100, gamma=0.001, kernel=sigmoid;, score=0.645 total time= 0.1s [CV 3/5] END C=100, gamma=0.001, kernel=sigmoid;, score=0.632 total time= 0.1s [CV 4/5] END C=100, gamma=0.001, kernel=sigmoid;, score=0.632 total time= 0.1s [CV 5/5] END C=100, gamma=0.001, kernel=sigmoid;, score=0.637 total time= 0.1s
GridSearchCV(estimator=SVC(class_weight='balanced'),
param_grid={'C': [0.1, 1, 10, 100], 'gamma': [1, 0.1, 0.01, 0.001],
'kernel': ['rbf', 'sigmoid']},
verbose=3)In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. GridSearchCV(estimator=SVC(class_weight='balanced'),
param_grid={'C': [0.1, 1, 10, 100], 'gamma': [1, 0.1, 0.01, 0.001],
'kernel': ['rbf', 'sigmoid']},
verbose=3)SVC(class_weight='balanced')
SVC(class_weight='balanced')
# training set
y_pred_svc_tr555 = SVC_grid.predict(X_train5)
accuracy_svc_tr555 = accuracy_score(y_train5, y_pred_svc_tr555)
print("Support Vector Classification Model Score (Training Set): %.2f%%" % (accuracy_svc_tr555 * 100.0))
# testing set
y_pred_svc_t555 = SVC_grid.predict(X_test5)
predictions_svc555 = [round(value) for value in y_pred_svc_t555]
# evaluate predictions
accuracy_svc_t555 = accuracy_score(y_test5, predictions_svc555)
print("Support Vector Classification Model Score (Testing Set): %.2f%%" % (accuracy_svc_t555 * 100.0))
Support Vector Classification Model Score (Training Set): 71.89% Support Vector Classification Model Score (Testing Set): 74.65%
plot_confusion_matrix(SVC_grid, X_test5, y_test5,cmap="YlGnBu");
# Printing the classification report
print(classification_report(y_test5, y_pred_svc_t555))
precision recall f1-score support
0 0.76 0.88 0.82 318
1 0.70 0.51 0.59 179
accuracy 0.75 497
macro avg 0.73 0.70 0.70 497
weighted avg 0.74 0.75 0.74 497
# Accuracy Score for each company.
SVC_accuracy3 = {'Aramco': accuracy_svc_t111*100, 'SARCO': accuracy_svc_t222*100, 'Rabigh': accuracy_svc_t333*100,\
'Bahri':accuracy_svc_t444*100, 'Aldrees': accuracy_svc_t555*100}
SVC_names3 = list(SVC_accuracy3.keys())
SVC_values3 = list(SVC_accuracy3.values())
fig, ax = plt.subplots(figsize=(15,8))
plt.title('Support Vector Classification using GridSearch CV Classification Accuracy Score per Company')
SVC_bars3 = ax.barh(SVC_names3, SVC_values3)
ax.bar_label(SVC_bars3, padding=5)
plt.xlim([0,100])
plt.show()
df_SVC = pd.DataFrame({
'Company': svc_names,
'SVC': svc_values,
'SVC_GridSearch': SVC_values3,
'SVC_RanomizedSearch':svc_values2
})
df_SVC
| Company | SVC | SVC_GridSearch | SVC_RanomizedSearch | |
|---|---|---|---|---|
| 0 | Aramco | 67.910448 | 63.432836 | 64.925373 |
| 1 | SARCO | 71.428571 | 90.342052 | 71.830986 |
| 2 | Rabigh | 56.740443 | 93.360161 | 56.136821 |
| 3 | Bahri | 62.374245 | 65.995976 | 62.374245 |
| 4 | Aldrees | 65.995976 | 74.647887 | 66.800805 |
df_SVC.to_csv('accuracy_SVC.csv',index = False)
# Calculate log loss (error) for all the Support Vector Classification Models
svc_log_loss = {
'Aramco': log_loss(y_true=y_test1, y_pred=y_pred_svc_t1, eps=1e-15, normalize=True, sample_weight=None, labels=None),
'SARCO': log_loss(y_true=y_test2, y_pred=y_pred_svc_t2, eps=1e-15, normalize=True, sample_weight=None, labels=None),
'Rabigh': log_loss(y_true=y_test3, y_pred=y_pred_svc_t3, eps=1e-15, normalize=True, sample_weight=None, labels=None),
'Bahri': log_loss(y_true=y_test4, y_pred=y_pred_svc_t4, eps=1e-15, normalize=True, sample_weight=None, labels=None),
'Aldrees': log_loss(y_true=y_test5, y_pred=y_pred_svc_t5, eps=1e-15, normalize=True, sample_weight=None, labels=None)
}
svc_log_loss_values = list(svc_log_loss.values())
svc_gridsearch_log_loss = {
'Aramco': log_loss(y_true=y_test1, y_pred=y_pred_svc_t111, eps=1e-15, normalize=True, sample_weight=None, labels=None),
'SARCO': log_loss(y_true=y_test2, y_pred=y_pred_svc_t222, eps=1e-15, normalize=True, sample_weight=None, labels=None),
'Rabigh': log_loss(y_true=y_test3, y_pred=y_pred_svc_t333, eps=1e-15, normalize=True, sample_weight=None, labels=None),
'Bahri': log_loss(y_true=y_test4, y_pred=y_pred_svc_t444, eps=1e-15, normalize=True, sample_weight=None, labels=None),
'Aldrees': log_loss(y_true=y_test5, y_pred=y_pred_svc_t555, eps=1e-15, normalize=True, sample_weight=None, labels=None)
}
svc_gridsearch_log_loss_values = list(svc_gridsearch_log_loss.values())
svc_randomsearch_log_loss = {
'Aramco': log_loss(y_true=y_test1, y_pred=y_pred_svc_t11, eps=1e-15, normalize=True, sample_weight=None, labels=None),
'SARCO': log_loss(y_true=y_test2, y_pred=y_pred_svc_t22, eps=1e-15, normalize=True, sample_weight=None, labels=None),
'Rabigh': log_loss(y_true=y_test3, y_pred=y_pred_svc_t33, eps=1e-15, normalize=True, sample_weight=None, labels=None),
'Bahri': log_loss(y_true=y_test4, y_pred=y_pred_svc_t44, eps=1e-15, normalize=True, sample_weight=None, labels=None),
'Aldrees': log_loss(y_true=y_test5, y_pred=y_pred_svc_t55, eps=1e-15, normalize=True, sample_weight=None, labels=None)
}
svc_randomsearch_log_loss_values = list(svc_randomsearch_log_loss.values())
df_svc_log_loss = pd.DataFrame({
'Company': svc_names,
'SVC': svc_log_loss_values,
'SVC_GridSearch': svc_gridsearch_log_loss_values,
'SVC_RanomizedSearch': svc_randomsearch_log_loss_values
})
df_svc_log_loss
| Company | SVC | SVC_GridSearch | SVC_RanomizedSearch | |
|---|---|---|---|---|
| 0 | Aramco | 11.083464 | 12.630078 | 12.114484 |
| 1 | SARCO | 9.868439 | 3.335780 | 9.729447 |
| 2 | Rabigh | 14.941333 | 2.293347 | 15.149817 |
| 3 | Bahri | 12.995578 | 11.744706 | 12.995578 |
| 4 | Aldrees | 11.744770 | 8.756372 | 11.466787 |
df_svc_log_loss.to_csv('error_SVC.csv',index = False)
svc_f1 = {
'Aramco': f1_score(y_true=y_test1, y_pred=y_pred_svc_t1, average='macro'),
'SARCO': f1_score(y_true=y_test2, y_pred=y_pred_svc_t2, average='macro'),
'Rabigh': f1_score(y_true=y_test3, y_pred=y_pred_svc_t3, average='macro'),
'Bahri': f1_score(y_true=y_test4, y_pred=y_pred_svc_t4, average='macro'),
'Aldrees': f1_score(y_true=y_test5, y_pred=y_pred_svc_t5, average='macro')
}
svc_f1_values = list(svc_f1.values())
svc_gridsearch_f1 = {
'Aramco': f1_score(y_true=y_test1, y_pred=y_pred_svc_t111, average='macro'),
'SARCO': f1_score(y_true=y_test2, y_pred=y_pred_svc_t222, average='macro'),
'Rabigh': f1_score(y_true=y_test3, y_pred=y_pred_svc_t333, average='macro'),
'Bahri': f1_score(y_true=y_test4, y_pred=y_pred_svc_t444, average='macro'),
'Aldrees': f1_score(y_true=y_test5, y_pred=y_pred_svc_t555, average='macro')
}
svc_gridsearch_f1_values = list(svc_gridsearch_f1.values())
svc_randomsearch_f1 = {
'Aramco': f1_score(y_true=y_test1, y_pred=y_pred_svc_t11, average='macro'),
'SARCO': f1_score(y_true=y_test2, y_pred=y_pred_svc_t22, average='macro'),
'Rabigh': f1_score(y_true=y_test3, y_pred=y_pred_svc_t33, average='macro'),
'Bahri': f1_score(y_true=y_test4, y_pred=y_pred_svc_t44, average='macro'),
'Aldrees': f1_score(y_true=y_test5, y_pred=y_pred_svc_t55, average='macro')
}
svc_randomsearch_f1_values = list(svc_randomsearch_f1.values())
df_svc_f1 = pd.DataFrame({
'Company': svc_names,
'SVC': svc_f1_values,
'SVC_GridSearch': svc_gridsearch_f1_values,
'SVC_RanomizedSearch': svc_randomsearch_f1_values
})
df_svc_f1
| Company | SVC | SVC_GridSearch | SVC_RanomizedSearch | |
|---|---|---|---|---|
| 0 | Aramco | 0.667532 | 0.566515 | 0.636605 |
| 1 | SARCO | 0.530408 | 0.599194 | 0.533145 |
| 2 | Rabigh | 0.400919 | 0.482830 | 0.397837 |
| 3 | Bahri | 0.623688 | 0.656758 | 0.623688 |
| 4 | Aldrees | 0.654226 | 0.704669 | 0.662048 |
df_svc_f1.to_csv('f1_SVC.csv',index = False)
# split the dataset into Training and Testing sets
X_trainlr1, X_testlr1, y_trainlr1, y_testlr1 = split_data(df_Aramco.drop('Volume', axis=1), 'Target')
#model
logre_aramco = LogisticRegression(class_weight="balanced").fit(X_trainlr1, y_trainlr1)
# predction:
logre_aramco_pred = logre_aramco.predict(X_testlr1)
#evalute
plot_confusion_matrix(logre_aramco, X_testlr1, y_testlr1,cmap="YlGnBu")
print(classification_report(y_testlr1,logre_aramco_pred))
accuracy_logre_t1 = accuracy_score(y_testlr1, logre_aramco_pred)
print("Logistic Regression Model Score (Testing Set): %.2f%%" % (accuracy_logre_t1 * 100.0))
precision recall f1-score support
0 0.53 0.57 0.55 54
1 0.70 0.66 0.68 80
accuracy 0.63 134
macro avg 0.62 0.62 0.62 134
weighted avg 0.63 0.63 0.63 134
Logistic Regression Model Score (Testing Set): 62.69%
# split the dataset into Training and Testing sets
X_trainlr2, X_testlr2, y_trainlr2, y_testlr2 = split_data(df_SARCO.drop('Volume', axis=1), 'Target')
#model
logre_sarco = LogisticRegression(class_weight="balanced").fit(X_trainlr2, y_trainlr2)
# predction:
logre_sarco_pred = logre_sarco.predict(X_testlr2)
#evalute
plot_confusion_matrix(logre_sarco, X_testlr2, y_testlr2,cmap="YlGnBu")
print(classification_report(y_testlr2,logre_sarco_pred))
accuracy_logre_t2 = accuracy_score(y_testlr2, logre_sarco_pred)
print("Logistic Regression Model Score (Testing Set): %.2f%%" % (accuracy_logre_t2 * 100.0))
precision recall f1-score support
0 0.98 0.73 0.84 468
1 0.14 0.72 0.24 29
accuracy 0.73 497
macro avg 0.56 0.73 0.54 497
weighted avg 0.93 0.73 0.80 497
Logistic Regression Model Score (Testing Set): 73.04%
# split the dataset into Training and Testing sets
X_trainlr3, X_testlr3, y_trainlr3, y_testlr3 = split_data(df_Rabigh.drop('Volume', axis=1), 'Target')
#model
logre_Rabigh=LogisticRegression(class_weight="balanced").fit(X_trainlr3, y_trainlr3)
# predction:
logre_Rabigh_pred = logre_Rabigh.predict(X_testlr3)
#evalute
plot_confusion_matrix(logre_Rabigh, X_testlr3, y_testlr3,cmap="YlGnBu")
print(classification_report(y_testlr3,logre_Rabigh_pred))
accuracy_logre_t3 = accuracy_score(y_testlr3, logre_Rabigh_pred)
print("Logistic Regression Model Score (Testing Set): %.2f%%" % (accuracy_logre_t3 * 100.0))
precision recall f1-score support
0 0.07 0.76 0.12 17
1 0.99 0.61 0.76 480
accuracy 0.62 497
macro avg 0.53 0.69 0.44 497
weighted avg 0.96 0.62 0.74 497
Logistic Regression Model Score (Testing Set): 61.97%
# split the dataset into Training and Testing sets
X_trainlr4, X_testlr4, y_trainlr4, y_testlr4 = split_data(df_Bahri.drop('Volume', axis=1), 'Target')
#model
logre_bahri= LogisticRegression(class_weight="balanced").fit(X_trainlr4, y_trainlr4)
# predction:
logre_bahri_pred = logre_bahri.predict(X_testlr4)
#evalute
plot_confusion_matrix(logre_bahri, X_testlr4, y_testlr4,cmap="YlGnBu")
print(classification_report(y_testlr4,logre_bahri_pred))
accuracy_logre_t4 = accuracy_score(y_testlr4, logre_bahri_pred )
print("Logistic Regression Model Score (Testing Set): %.2f%%" % (accuracy_logre_t4 * 100.0))
precision recall f1-score support
0 0.57 0.69 0.62 222
1 0.70 0.58 0.63 275
accuracy 0.63 497
macro avg 0.63 0.64 0.63 497
weighted avg 0.64 0.63 0.63 497
Logistic Regression Model Score (Testing Set): 62.98%
# split the dataset into Training and Testing sets
X_trainlr5, X_testlr5, y_trainlr5, y_testlr5 = split_data(df_Aldrees.drop('Volume', axis=1), 'Target')
#model
logre_Aldrees= LogisticRegression(class_weight="balanced").fit(X_trainlr5, y_trainlr5)
# predction:
logre_Aldrees_pred = logre_Aldrees.predict(X_testlr5)
#evalute
plot_confusion_matrix(logre_Aldrees, X_testlr5, y_testlr5,cmap="YlGnBu")
print(classification_report(y_testlr5,logre_Aldrees_pred))
accuracy_logre_t5 = accuracy_score(y_testlr5, logre_Aldrees_pred)
print("Logistic Regression Model Score (Testing Set): %.2f%%" % (accuracy_logre_t5 * 100.0))
precision recall f1-score support
0 0.80 0.66 0.72 318
1 0.54 0.70 0.61 179
accuracy 0.68 497
macro avg 0.67 0.68 0.67 497
weighted avg 0.70 0.68 0.68 497
Logistic Regression Model Score (Testing Set): 67.61%
Logistic Regression Overview (Accuracy,log_loss, F1 score)
# Accuracy Score for each company.
logre_accuracy = {'Aramco': accuracy_logre_t1*100, 'SARCO': accuracy_logre_t2*100, 'Rabigh': accuracy_logre_t3*100, 'Bahri':accuracy_logre_t4*100, 'Aldrees': accuracy_logre_t5*100}
print(logre_accuracy)
logre_names = list(logre_accuracy.keys())
logre_values = list(logre_accuracy.values())
#create accuracy plot for each company
fig, ax = plt.subplots(figsize=(15,8))
plt.title('Logistic Regression Accuracy Score per Company')
bars = ax.barh(logre_names,logre_values)
ax.bar_label(bars, fmt='%.2f%%')
plt.show()
{'Aramco': 62.68656716417911, 'SARCO': 73.03822937625755, 'Rabigh': 61.97183098591549, 'Bahri': 62.97786720321932, 'Aldrees': 67.6056338028169}
#use function to create accuracy Datafarme for each company
df_logre = pd.DataFrame({
'Company': logre_names,
'Logistic Regression':logre_values,
})
#call function
df_logre
#save dataframe in csv
#df_logre.to_csv('accuracy_logre.csv',index = False)
| Company | Logistic Regression | |
|---|---|---|
| 0 | Aramco | 62.686567 |
| 1 | SARCO | 73.038229 |
| 2 | Rabigh | 61.971831 |
| 3 | Bahri | 62.977867 |
| 4 | Aldrees | 67.605634 |
# Calculate log loss (error)
logre_log_loss = {
'Aramco': log_loss(y_true=y_testlr1, y_pred=logre_aramco_pred, eps=1e-15, normalize=True, sample_weight=None, labels=None),
'SARCO': log_loss(y_true=y_testlr2, y_pred=logre_sarco_pred, eps=1e-15, normalize=True, sample_weight=None, labels=None),
'Rabigh': log_loss(y_true=y_testlr3, y_pred=logre_Rabigh_pred, eps=1e-15, normalize=True, sample_weight=None, labels=None),
'Bahri': log_loss(y_true=y_testlr4, y_pred=logre_bahri_pred, eps=1e-15, normalize=True, sample_weight=None, labels=None),
'Aldrees': log_loss(y_true=y_testlr5, y_pred=logre_Aldrees_pred, eps=1e-15, normalize=True, sample_weight=None, labels=None)
}
#store value in list
logre_log_loss_values= list(logre_log_loss.values())
#use function to create loss error Datafarme for each company
df_logre_log_loss = pd.DataFrame({
'Company': logre_names,
'Logistic Regression': logre_log_loss_values,
})
#call function
df_logre_log_loss
#save dataframe in csv
df_logre_log_loss.to_csv('logre_log_loss.csv')
# Calculate f1 score
logre_f1_score = {
'Aramco': f1_score(y_true=y_testlr1, y_pred=logre_aramco_pred),
'SARCO': f1_score(y_true=y_testlr2, y_pred=logre_sarco_pred),
'Rabigh': f1_score(y_true=y_testlr3, y_pred=logre_Rabigh_pred),
'Bahri': f1_score(y_true=y_testlr4, y_pred=logre_bahri_pred),
'Aldrees': f1_score(y_true=y_testlr5, y_pred=logre_Aldrees_pred,)
}
#store value in list
logre_f1_score_values = list(logre_f1_score.values())
#use function to create f1 score Datafarme for each company
df_logre_f1_score = pd.DataFrame({
'Company': logre_names,
'Logistic Regression': logre_f1_score_values,
})
#call function
df_logre_f1_score
#save dataframe in csv
df_logre_f1_score.to_csv('logre_f1_score.csv')
full_data_with_allFeatures = pd.read_csv('Data-Science-Campus-Capstone-Project/Data2/cleaned_data_frame.csv')
df_Aramco=full_data_with_allFeatures[full_data_with_allFeatures['Name']=='Saudi Aramco']
df_SARCO=full_data_with_allFeatures[full_data_with_allFeatures['Name']=='SARCO']
df_Rabigh=full_data_with_allFeatures[full_data_with_allFeatures['Name']=='Petro Rabigh']
df_Bahri=full_data_with_allFeatures[full_data_with_allFeatures['Name']=='Bahri']
df_Aldrees=full_data_with_allFeatures[full_data_with_allFeatures['Name']=='Aldrees']
column_names = [
'Volume','EMA_9',
'SMA_5',
'SMA_10',
'SMA_15',
'SMA_20',
'SMA_25',
'SMA_30',
'MACD',
'RSI',
'Stochastic',
'Open-Close',
'High-Low']
column_names.append('Target')
# here we use Burta as intial and modfy it to get the best result
Burta_column_names = ['SMA_5', 'MACD', 'RSI']
Burta_column_names.append('Target')
data_Aramco = df_Aramco[Burta_column_names]
window_length = 7 # window_length is usually selected based on prior domain knowledge or by trial and error
stride = None # None for non-overlapping (stride = window_length) (default = 1). This depends on how often you want to predict once the model is trained
start = 0 # use all data since the first time stamp (default = 0)
get_x = data_Aramco.columns[:-1] # Indicates which are the columns that contain the x data.
get_y = 'Target' # In multivariate time series, you must indicate which is/are the y columns
horizon = 0 # 0 means y is taken from the last time stamp of the time sequence (default = 0)
seq_first = True
X, y = SlidingWindow(window_length, get_x=get_x, get_y=get_y)(data_Aramco)
!python -m pip uninstall matplotlib
!pip install matplotlib==3.1.3
splits = get_splits(y, valid_size=.2, stratify=True, random_state=42, shuffle=False)
#3d tensor of shape (samples, vars, len),
tfms = [None, [Categorize()]]
batch_tfms = [TSStandardize(by_sample=True)]
dls_Aramco = get_ts_dls(X, y, splits=splits, tfms=tfms, drop_last=False, shuffle_train=False, batch_tfms=batch_tfms, bs=128)
model_Aramco = build_ts_model(ROCKET,kss=np.array([6, 9, 11]), dls=dls) # n_kernels=10_000, kss=[7, 9, 11] set by default, but you can pass other values as kwargs
X_train_Aramco, y_train_Aramco = create_rocket_features(dls_Aramco.train, model_Aramco)
X_test_Aramco, y_test_Aramco = create_rocket_features(dls_Aramco.valid, model_Aramco)
X_train_Aramco.shape, X_test_Aramco.shape
((531, 20000), (132, 20000))
Ridge Classifer
ridge_Aramco = RidgeClassifierCV(alphas=np.logspace(-2, 2, 100), normalize=True,class_weight='balanced')
ridge_Aramco.fit(X_train_Aramco, y_train_Aramco)
print(f'alpha: {ridge_Aramco.alpha_:.2E} train: {ridge_Aramco.score(X_train_Aramco, y_train_Aramco):.5f} test: {ridge_Aramco.score(X_test_Aramco, y_test_Aramco):.5f}')
alpha: 1.00E+02 train: 0.70810 test: 0.56061
y_pred_Aramco = ridge_Aramco.predict(X_test_Aramco)
log_loss(y_test_Aramco,y_pred_Aramco)
15.17631680625118
plot_confusion_matrix(ridge_Aramco, X_test_Aramco, y_test_Aramco,cmap="YlGnBu");
# Printing the classification report
print(classification_report(y_test_Aramco, y_pred_Aramco))
precision recall f1-score support
0 0.56 0.52 0.54 65
1 0.56 0.60 0.58 67
accuracy 0.56 132
macro avg 0.56 0.56 0.56 132
weighted avg 0.56 0.56 0.56 132
data_SARCO_Burta = df_SARCO[Burta_column_names]
data_SARCO = df_SARCO[column_names]
window_length = 7 # window_length is usually selected based on prior domain knowledge or by trial and error
stride = None # None for non-overlapping (stride = window_length) (default = 1). This depends on how often you want to predict once the model is trained
start = 0 # use all data since the first time stamp (default = 0)
get_x = data_SARCO_Burta.columns[:-1] # Indicates which are the columns that contain the x data.
get_y = 'Target' # In multivariate time series, you must indicate which is/are the y columns
horizon = 0 # 0 means y is taken from the last time stamp of the time sequence (default = 0)
seq_first = True
X, y = SlidingWindow(window_length, get_x=get_x, get_y=get_y)(data_SARCO_Burta)
splits = get_splits(y, valid_size=.2, stratify=True, random_state=42, shuffle=False)
#3d tensor of shape (samples, vars, len),
tfms = [None, [Categorize()]]
batch_tfms = [TSStandardize(by_sample=True)]
dls_SARCO = get_ts_dls(X, y, splits=splits, tfms=tfms, drop_last=False, shuffle_train=False, batch_tfms=batch_tfms, bs=64)
model_SARCO = build_ts_model(ROCKET,kss=np.array([6, 9, 11]), dls=dls_SARCO) # n_kernels=10_000, kss=[7, 9, 11] set by default, but you can pass other values as kwargs
X_train_SARCO, y_train_SARCO = create_rocket_features(dls_SARCO.train, model_SARCO)
X_test_SARCO, y_test_SARCO = create_rocket_features(dls_SARCO.valid, model_SARCO)
X_train_SARCO.shape, X_test_SARCO.shape
((1980, 20000), (494, 20000))
Ridge Classifer
ridge_SARCO = RidgeClassifierCV(alphas=np.logspace(-2, 2, 100), normalize=True,class_weight='balanced')
ridge_SARCO.fit(X_train_SARCO, y_train_SARCO)
print(f'alpha: {ridge_SARCO.alpha_:.2E} train: {ridge_SARCO.score(X_train_SARCO, y_train_SARCO):.5f} test: {ridge_SARCO.score(X_test_SARCO, y_test_SARCO):.5f}')
alpha: 4.75E+01 train: 0.76414 test: 0.96964
y_pred_SARCO = ridge_SARCO.predict(X_test_SARCO)
log_loss(y_test_SARCO,y_pred_SARCO)
1.048759601428486
plot_confusion_matrix(ridge_SARCO, X_test_SARCO, y_test_SARCO,cmap="YlGnBu");
print(classification_report(y_test_SARCO, y_pred_SARCO))
precision recall f1-score support
0 0.98 0.99 0.98 486
1 0.00 0.00 0.00 8
accuracy 0.97 494
macro avg 0.49 0.49 0.49 494
weighted avg 0.97 0.97 0.97 494
df_Rabigh_Burta = df_Rabigh[Burta_column_names]
data_Rabigh = df_Rabigh[column_names]
window_length = 7 # window_length is usually selected based on prior domain knowledge or by trial and error
stride = None # None for non-overlapping (stride = window_length) (default = 1). This depends on how often you want to predict once the model is trained
start = 0 # use all data since the first time stamp (default = 0)
get_x = df_Rabigh_Burta.columns[:-1] # Indicates which are the columns that contain the x data.
get_y = 'Target' # In multivariate time series, you must indicate which is/are the y columns
horizon = 0 # 0 means y is taken from the last time stamp of the time sequence (default = 0)
seq_first = True
X, y = SlidingWindow(window_length, get_x=get_x, get_y=get_y)(df_Rabigh_Burta)
splits = get_splits(y, valid_size=.2, stratify=True, random_state=42, shuffle=False)
#3d tensor of shape (samples, vars, len),
tfms = [None, [Categorize()]]
batch_tfms = [TSStandardize(by_sample=True)]
dls_Rabigh = get_ts_dls(X, y, splits=splits, tfms=tfms, drop_last=False, shuffle_train=False, batch_tfms=batch_tfms, bs=128)
model_Rabigh = build_ts_model(ROCKET,kss=np.array([6, 9, 11]), dls=dls_Rabigh) # n_kernels=10_000, kss=[7, 9, 11] set by default, but you can pass other values as kwargs
X_train_Rabigh, y_train_Rabigh = create_rocket_features(dls_Rabigh.train, model_Rabigh)
X_test_Rabigh, y_test_Rabigh = create_rocket_features(dls_Rabigh.valid, model_Rabigh)
X_train_Rabigh.shape, X_test_Rabigh.shape
((1980, 20000), (495, 20000))
Ridge Classifer
ridge_Rabigh = RidgeClassifierCV(alphas=np.logspace(-2, 2, 100), normalize=True,class_weight='balanced')
ridge_Rabigh.fit(X_train_Rabigh, y_train_Rabigh)
print(f'alpha: {ridge_Rabigh.alpha_:.2E} train: {ridge_Rabigh.score(X_train_Rabigh, y_train_Rabigh):.5f} test: {ridge_Rabigh.score(X_test_Rabigh, y_test_Rabigh):.5f}')
alpha: 1.00E+02 train: 0.75758 test: 0.87677
y_pred_Rabigh = ridge_Rabigh.predict(X_test_Rabigh)
log_loss(y_test_Rabigh,y_pred_Rabigh)
4.256309810230007
plot_confusion_matrix(ridge_Rabigh, X_test_Rabigh, y_test_Rabigh,cmap="YlGnBu");
print(classification_report(y_test_Rabigh, y_pred_Rabigh))
precision recall f1-score support
0 0.02 0.09 0.03 11
1 0.98 0.89 0.93 484
accuracy 0.88 495
macro avg 0.50 0.49 0.48 495
weighted avg 0.96 0.88 0.91 495
#df_Bahri
df_Bahri_Burta = df_Bahri[Burta_column_names]
data_Bahri = df_Bahri[column_names]
window_length = 7 # window_length is usually selected based on prior domain knowledge or by trial and error
stride = None # None for non-overlapping (stride = window_length) (default = 1). This depends on how often you want to predict once the model is trained
start = 0 # use all data since the first time stamp (default = 0)
get_x = df_Bahri_Burta.columns[:-1] # Indicates which are the columns that contain the x data.
get_y = 'Target' # In multivariate time series, you must indicate which is/are the y columns
horizon = 0 # 0 means y is taken from the last time stamp of the time sequence (default = 0)
seq_first = True
X, y = SlidingWindow(window_length, get_x=get_x, get_y=get_y)(df_Bahri_Burta)
splits = get_splits(y, valid_size=.2, stratify=True, random_state=42, shuffle=False)
#3d tensor of shape (samples, vars, len),
tfms = [None, [Categorize()]]
batch_tfms = [TSStandardize(by_sample=True)]
dls_Bahri = get_ts_dls(X, y, splits=splits, tfms=tfms, drop_last=False, shuffle_train=False, batch_tfms=batch_tfms, bs=64)
model_Bahri = build_ts_model(ROCKET,kss=np.array([6, 9, 11]), dls=dls_Bahri) # n_kernels=10_000, kss=[7, 9, 11] set by default, but you can pass other values as kwargs
X_train_Bahri, y_train_Bahri = create_rocket_features(dls_Bahri.train, model_Bahri)
X_test_Bahri, y_test_Bahri = create_rocket_features(dls_Bahri.valid, model_Bahri)
X_train_Bahri.shape, X_test_Bahri.shape
((1980, 20000), (494, 20000))
Ridge classifer
ridge_Bahri = RidgeClassifierCV(alphas=np.logspace(-1, 1, 100), normalize=True,class_weight='balanced')
ridge_Bahri.fit(X_train_Bahri, y_train_Bahri)
print(f'alpha: {ridge_Bahri.alpha_:.2E} train: {ridge_Bahri.score(X_train_Bahri, y_train_Bahri):.5f} test: {ridge_Bahri.score(X_test_Bahri, y_test_Bahri):.5f}')
alpha: 1.00E+01 train: 0.71111 test: 0.71255
y_pred_Bahri = ridge_Bahri.predict(X_test_Bahri)
log_loss(y_test_Bahri,y_pred_Bahri)
9.928352627036624
plot_confusion_matrix(ridge_Bahri, X_test_Bahri, y_test_Bahri,cmap="YlGnBu");
print(classification_report(y_test_Bahri, y_pred_Bahri))
precision recall f1-score support
0 0.23 0.04 0.07 130
1 0.74 0.95 0.83 364
accuracy 0.71 494
macro avg 0.48 0.50 0.45 494
weighted avg 0.60 0.71 0.63 494
df_Aldrees_Burta = df_Aldrees[Burta_column_names]
data_Aldrees = df_Aldrees[column_names]
window_length = 7 # window_length is usually selected based on prior domain knowledge or by trial and error
stride = None # None for non-overlapping (stride = window_length) (default = 1). This depends on how often you want to predict once the model is trained
start = 0 # use all data since the first time stamp (default = 0)
get_x = df_Aldrees_Burta.columns[:-1] # Indicates which are the columns that contain the x data.
get_y = 'Target' # In multivariate time series, you must indicate which is/are the y columns
horizon = 0 # 0 means y is taken from the last time stamp of the time sequence (default = 0)
seq_first = True
X, y = SlidingWindow(window_length, get_x=get_x, get_y=get_y)(df_Aldrees_Burta)
splits = get_splits(y, valid_size=.2, stratify=True, random_state=42, shuffle=False)
#3d tensor of shape (samples, vars, len),
tfms = [None, [Categorize()]]
batch_tfms = [TSStandardize(by_sample=True)]
dls_Aldrees = get_ts_dls(X, y, splits=splits, tfms=tfms, drop_last=False, shuffle_train=False, batch_tfms=batch_tfms, bs=64)
model_Aldrees = build_ts_model(ROCKET,kss=np.array([6, 9, 11]), dls=dls_Aldrees) # n_kernels=10_000, kss=[7, 9, 11] set by default, but you can pass other values as kwargs
X_train_Aldrees, y_train_Aldrees = create_rocket_features(dls_Aldrees.train, model_Aldrees)
X_test_Aldrees, y_test_Aldrees = create_rocket_features(dls_Aldrees.valid, model_Aldrees)
X_train_Aldrees.shape, X_test_Aldrees.shape
((1980, 20000), (495, 20000))
RidgeCV
ridge_Aldrees = RidgeClassifierCV(alphas=np.logspace(-2, 2, 100), normalize=True,class_weight='balanced')
ridge_Aldrees.fit(X_train_Aldrees, y_train_Aldrees)
print(f'alpha: {ridge_Aldrees.alpha_:.2E} train: {ridge_Aldrees.score(X_train_Aldrees, y_train_Aldrees):.5f} test: {ridge_Aldrees.score(X_test_Aldrees, y_test_Aldrees):.5f}')
alpha: 8.30E+01 train: 0.63990 test: 0.62222
y_pred_Aldrees = ridge_Aldrees.predict(X_test_Aldrees)
log_loss(y_test_Aldrees,y_pred_Aldrees)
13.048150189860683
plot_confusion_matrix(ridge_Aldrees, X_test_Aldrees, y_test_Aldrees,cmap="YlGnBu");
print(classification_report(y_test_Aldrees, y_pred_Aldrees))
precision recall f1-score support
0 0.77 0.73 0.75 384
1 0.21 0.25 0.23 111
accuracy 0.62 495
macro avg 0.49 0.49 0.49 495
weighted avg 0.65 0.62 0.63 495
compNames = ['Aramco','SARCO','Rabigh','Bahri','Aldrees']
#Acc
#y_test_Aramco, y_pred_Aramco
#log_loss(y_true=y_test_Aramco, y_pred=y_pred_Aramco, normalize=True)
#accuracy_score(y_true=y_test_Aramco, y_pred=y_pred_Aramco)
#_Aldrees
results_dic = {'Aramco':[{'ROCKET + RidgeClassifierCV':accuracy_score(y_true=y_test_Aramco, y_pred=y_pred_Aramco)}],
'SARCO':[{'ROCKET + RidgeClassifierCV':accuracy_score(y_true=y_test_SARCO, y_pred=y_pred_SARCO)}],
'Rabigh':[{'ROCKET + RidgeClassifierCV':accuracy_score(y_true=y_test_Rabigh, y_pred=y_pred_Rabigh)}],
'Bahri':[{'ROCKET + RidgeClassifierCV':accuracy_score(y_true=y_test_Bahri, y_pred=y_pred_Bahri)}],
'Aldrees':[{'ROCKET + RidgeClassifierCV':accuracy_score(y_true=y_test_Aldrees, y_pred=y_pred_Aldrees)}]}
#f1
results_dic_f1 = {'Aramco':[{'ROCKET + RidgeClassifierCV':f1_score(y_true=y_test_Aramco, y_pred=y_pred_Aramco, average='macro')}],
'SARCO':[{'ROCKET + RidgeClassifierCV':f1_score(y_true=y_test_SARCO, y_pred=y_pred_SARCO, average='macro')}],
'Rabigh':[{'ROCKET + RidgeClassifierCV':f1_score(y_true=y_test_Rabigh, y_pred=y_pred_Rabigh, average='macro')}],
'Bahri':[{'ROCKET + RidgeClassifierCV':f1_score(y_true=y_test_Bahri, y_pred=y_pred_Bahri, average='macro')}],
'Aldrees':[{'ROCKET + RidgeClassifierCV':f1_score(y_true=y_test_Aramco, y_pred=y_pred_Aramco, average='macro')}]}
#loss
errors_dic = {'Aramco':[{'ROCKET + RidgeClassifierCV':log_loss(y_true=y_test_Aramco, y_pred=y_pred_Aramco, normalize=True)}],
'SARCO':[{'ROCKET + RidgeClassifierCV':log_loss(y_true=y_test_SARCO, y_pred=y_pred_SARCO, normalize=True)}],
'Rabigh':[{'ROCKET + RidgeClassifierCV':log_loss(y_true=y_test_Rabigh, y_pred=y_pred_Rabigh, normalize=True)}],
'Bahri':[{'ROCKET + RidgeClassifierCV':log_loss(y_true=y_test_Bahri, y_pred=y_pred_Bahri, normalize=True)}],
'Aldrees':[{'ROCKET + RidgeClassifierCV':log_loss(y_true=y_test_Aramco, y_pred=y_pred_Aramco, normalize=True)}]}
# RidgeClassifierCVAll_features = {
# 'Aramco': round(results_dic['Aramco'][0]['RidgeClassifierCV + All features']*100,2),
# 'SARCO': round(results_dic['SARCO'][0]['RidgeClassifierCV + All features']*100,2),
# 'Rabigh': round(results_dic['Rabigh'][0]['RidgeClassifierCV + All features']*100,2),
# 'Bahri': round(results_dic['Bahri'][0]['RidgeClassifierCV + All features']*100,2),
# 'Aldrees':round(results_dic['Aldrees'][0]['RidgeClassifierCV + All features']*100,2)
# }
# RidgeClassifierCVAll_features_values = list(RidgeClassifierCVAll_features.values())
RidgeClassifierCVborutaSelected_acc = {
'Aramco': round(results_dic['Aramco'][0]['ROCKET + RidgeClassifierCV']*100,2),
'SARCO': round(results_dic['SARCO'][0]['ROCKET + RidgeClassifierCV']*100,2),
'Rabigh': round(results_dic['Rabigh'][0]['ROCKET + RidgeClassifierCV']*100,2),
'Bahri': round(results_dic['Bahri'][0]['ROCKET + RidgeClassifierCV']*100,2),
'Aldrees': round(results_dic['Aldrees'][0]['ROCKET + RidgeClassifierCV']*100,2)
}
RidgeClassifierCVborutaSelected_acc = list(RidgeClassifierCVborutaSelected_acc.values())
df_ROKET = pd.DataFrame({
'Company': compNames,
#'ROCKET+RidgeClassifierCV + All Features': RidgeClassifierCVAll_features_values,
'ROCKET + RidgeClassifierCV': RidgeClassifierCVborutaSelected_acc
})
df_ROKET
| Company | ROCKET + RidgeClassifierCV | |
|---|---|---|
| 0 | Aramco | 56.06 |
| 1 | SARCO | 96.96 |
| 2 | Rabigh | 87.68 |
| 3 | Bahri | 71.26 |
| 4 | Aldrees | 62.22 |
df_ROKET.to_csv('Accurecies_results.csv')
# RidgeClassifierCVAll_features = {
# 'Aramco': round(results_dic_f1['Aramco'][0]['RidgeClassifierCV + All features']*100,2),
# 'SARCO': round(results_dic_f1['SARCO'][0]['RidgeClassifierCV + All features']*100,2),
# 'Rabigh': round(results_dic_f1['Rabigh'][0]['RidgeClassifierCV + All features']*100,2),
# 'Bahri': round(results_dic_f1['Bahri'][0]['RidgeClassifierCV + All features']*100,2),
# 'Aldrees':round(results_dic_f1['Aldrees'][0]['RidgeClassifierCV + All features']*100,2)
# }
# RidgeClassifierCVAll_features_values = list(RidgeClassifierCVAll_features.values())
RidgeClassifierCVboruta_f1 = {
'Aramco': round(results_dic_f1['Aramco'][0]['ROCKET + RidgeClassifierCV']*100,2),
'SARCO': round(results_dic_f1['SARCO'][0]['ROCKET + RidgeClassifierCV']*100,2),
'Rabigh': round(results_dic_f1['Rabigh'][0]['ROCKET + RidgeClassifierCV']*100,2),
'Bahri': round(results_dic_f1['Bahri'][0]['ROCKET + RidgeClassifierCV']*100,2),
'Aldrees': round(results_dic_f1['Aldrees'][0]['ROCKET + RidgeClassifierCV']*100,2)
}
RidgeClassifierCVboruta_f1 = list(RidgeClassifierCVboruta_f1.values())
df_ROKET_f1 = pd.DataFrame({
'Company': compNames,
#'ROCKET+RidgeClassifierCV + All Features': RidgeClassifierCVAll_features_values,
'ROCKET + RidgeClassifierCV': RidgeClassifierCVboruta_f1
})
df_ROKET_f1
| Company | ROCKET + RidgeClassifierCV | |
|---|---|---|
| 0 | Aramco | 55.97 |
| 1 | SARCO | 49.23 |
| 2 | Rabigh | 48.30 |
| 3 | Bahri | 44.80 |
| 4 | Aldrees | 55.97 |
df_ROKET_f1.to_csv('f1_results.csv')
# RidgeClassifierCVAll_features = {
# 'Aramco': errors_dic['Aramco'][0]['RidgeClassifierCV + All features'],
# 'SARCO': errors_dic['SARCO'][0]['RidgeClassifierCV + All features'],
# 'Rabigh': errors_dic['Rabigh'][0]['RidgeClassifierCV + All features'],
# 'Bahri': errors_dic['Bahri'][0]['RidgeClassifierCV + All features'],
# 'Aldrees':errors_dic['Aldrees'][0]['RidgeClassifierCV + All features']
# }
# RidgeClassifierCVAll_features_values = list(RidgeClassifierCVAll_features.values())
RidgeClassifierCVboruta_errors = {
'Aramco': errors_dic['Aramco'][0]['ROCKET + RidgeClassifierCV'],
'SARCO': errors_dic['SARCO'][0]['ROCKET + RidgeClassifierCV'],
'Rabigh': errors_dic['Rabigh'][0]['ROCKET + RidgeClassifierCV'],
'Bahri': errors_dic['Bahri'][0]['ROCKET + RidgeClassifierCV'],
'Aldrees': errors_dic['Aldrees'][0]['ROCKET + RidgeClassifierCV']
}
RidgeClassifierCVboruta_errors = list(RidgeClassifierCVboruta_errors.values())
df_ROKET_errors = pd.DataFrame({
'Company': compNames,
#'ROCKET+RidgeClassifierCV + All Features': RidgeClassifierCVAll_features_values,
'ROCKET+RidgeClassifierCV + Boruta Selected Features': RidgeClassifierCVboruta_errors
})
df_ROKET_errors
| Company | ROCKET+RidgeClassifierCV + Boruta Selected Features | |
|---|---|---|
| 0 | Aramco | 15.176317 |
| 1 | SARCO | 1.048760 |
| 2 | Rabigh | 4.256310 |
| 3 | Bahri | 9.928353 |
| 4 | Aldrees | 15.176317 |
df_ROKET_errors.to_csv('errors.csv')
fig, ((ax, ax1,ax2,ax3,ax4))= plt.subplots(ncols=1,nrows=5, figsize=(20,30), )
#Best Accuracy Score of Each Model for Aramco
#1-get values
Compare_accuracy1 = {
'Logistic Regression': df_logre.iloc[0,1],
'XGBoost':df_xgb.iloc[0,1:].max(),
'Random Forest':df_RF.iloc[0,1:].max(),
'SVC': df_SVC.iloc[0,1:].max(),
'ROCKET+RidgeClassifierCV':df_ROKET.iloc[0,1]
}
compare_names = list(Compare_accuracy1.keys())
compare_values = list(Compare_accuracy1.values())
#2- create plot
ax.set_title('Best Model Accuracy of Each Model Type for Aramco ', fontsize=18)
bars = ax.barh(compare_names,compare_values)
ax.bar_label(bars, fmt='%.2f%%', fontsize=13 )
ax.set_xlim([0, 100])
#ax.set_ylabel('Model', fontsize=16)
ax.set_yticklabels(compare_names, fontsize=14)
#ax.set(xlabel=None)
#ax.set_tick_params(top='off', bottom='off', right='off', labelleft='on', labelbottom='off')
ax.spines[['right', 'top', 'left', 'bottom']].set_visible(False)
#ax.spines['top'].set_visible(False)
ax.set_xticks([])
#Best Accuracy Score of Each Model for sarco
#1-get values
Compare_accuracy2 = {
'Logistic Regression': df_logre.iloc[1,1],
'XGBoost':df_xgb.iloc[1,1:].max(),
'Random Forest':df_RF.iloc[1,1:].max(),
'SVC': df_SVC.iloc[1,1:].max(),
'ROCKET+RidgeClassifierCV':df_ROKET.iloc[1,1]
}
compare_names2 = list(Compare_accuracy2.keys())
compare_values2 = list(Compare_accuracy2.values())
#2-create plot
ax1.set_title('Best Model Accuracy of Each Model Type for SARCO', fontsize=18)
bars = ax1.barh(compare_names2,compare_values2)
ax1.bar_label(bars, fmt='%.2f%%', fontsize=13 )
ax1.set_xlim([0, 100])
ax1.set_yticklabels(compare_names, fontsize=14)
ax1.spines[['right', 'top', 'left', 'bottom']].set_visible(False)
ax1.set_xticks([])
#Best Accuracy Score of Each Model for Rabigh
#1-get values
Compare_accuracy3 = {
'Logistic Regression': df_logre.iloc[2,1],
'XGBoost':df_xgb.iloc[2,1:].max(),
'Random Forest':df_RF.iloc[2,1:].max(),
'SVC': df_SVC.iloc[2,1:].max(),
'ROCKET+RidgeClassifierCV':df_ROKET.iloc[2,1]
}
compare_names3 = list(Compare_accuracy3.keys())
compare_values3 = list(Compare_accuracy3.values())
#2- create plot
ax2.set_title('Best Model Accuracy of Each Model Type for Rabigh', fontsize=18)
bars = ax2.barh(compare_names3,compare_values3)
ax2.bar_label(bars, fmt='%.2f%%', fontsize=13)
ax2.set_xlim([0, 100])
ax2.set_yticklabels(compare_names, fontsize=14)
ax2.spines[['right', 'top', 'left', 'bottom']].set_visible(False)
ax2.set_xticks([])
#Best Accuracy Score of Each Model for Bahri
#1-get Values
Compare_accuracy4 = {
'Logistic Regression': df_logre.iloc[3,1],
'XGBoost':df_xgb.iloc[3,1:].max(),
'Random Forest':df_RF.iloc[3,1:].max(),
'SVC': df_SVC.iloc[3,1:].max(),
'ROCKET+RidgeClassifierCV':df_ROKET.iloc[3,1]
}
compare_names4 = list(Compare_accuracy4.keys())
compare_values4 = list(Compare_accuracy4.values())
#2-create Plot
ax3.set_title('Best Model Accuracy of Each Model Type for Bahri', fontsize=18)
bars = ax3.barh(compare_names4,compare_values4)
ax3.bar_label(bars, fmt='%.2f%%', fontsize=13)
ax3.set_xlim([0, 100])
ax3.set_yticklabels(compare_names, fontsize=14)
ax3.spines[['right', 'top', 'left', 'bottom']].set_visible(False)
ax3.set_xticks([])
#Best Accuracy Score of Each Model for Aldrees
#1-get Values
Compare_accuracy5 = {
'Logistic Regression': df_logre.iloc[4,1],
'XGBoost':df_xgb.iloc[4,1:].max(),
'Random Forest':df_RF.iloc[4,1:].max(),
'SVC': df_SVC.iloc[4,1:].max(),
'ROCKET+RidgeClassifierCV':df_ROKET.iloc[4,1]
}
compare_names5 = list(Compare_accuracy5.keys())
compare_values5 = list(Compare_accuracy5.values())
#2-create Plot
ax4.set_title('Best Model Accuracy of Each Model Type for Aldrees', fontsize=18)
bars = ax4.barh(compare_names5,compare_values5)
ax4.bar_label(bars, fmt='%.2f%%', fontsize=13)
ax4.set_xlim([0, 100])
ax4.set_yticklabels(compare_names, fontsize=14)
ax4.spines[['right', 'top', 'left', 'bottom']].set_visible(False)
ax4.set_xticks([])
plt.tight_layout()
plt.subplots_adjust(
hspace=0.17)
plt.savefig("figure.jpeg", dpi=1200) # save as png
plt.show()
BaseLine_result=[score1,score2,score3,score4,score5]
RandomForest_result=[df_RF.iloc[0,1:].max(),df_RF.iloc[1,1:].max(),df_RF.iloc[2,1:].max(),df_RF.iloc[3,1:].max(),df_RF.iloc[4,1:].max()]
x = np.arange(len(compNames)) # the label locations
width = 0.35 # the width of the bars
fig, ax = plt.subplots(figsize = (10, 5))
rects1 = ax.bar(x - width/2, BaseLine_result, width, label='BaseLine',color='#51d2bb')
rects2 = ax.bar(x + width/2, RandomForest_result, width, label='Random Forest',color='#116985')
# Add some text for labels, title and custom x-axis tick labels, etc.
ax.set_ylabel('Accurcy perstange')
ax.set_title('Baseline Vs Random Forest')
ax.set_xticks(x)
ax.set_xticklabels(compNames)
ax.legend()
ax.bar_label(rects1,fmt='%1.2f%%', padding=3)
ax.bar_label(rects2,fmt='%1.2f%%', padding=3)
fig.tight_layout()
plt.savefig("figure.jpeg", dpi=1200) # save as png
plt.show()
RF_clf= RandomForestClassifier(class_weight="balanced", random_state=42)
params_grid = {
'bootstrap': [True],
'max_depth': [80],
'max_features': [3,5,8,7, 10,13],
'min_samples_leaf': [3, 5],
'min_samples_split': [2, 6, 8],
'n_estimators': [200,400]
}
RF_grid = GridSearchCV(
RF_clf,
params_grid,
cv = 10,
n_jobs=-1,
verbose=2);
X_train1, X_test1, y_train1, y_test1 = split_data(df_Aramco, 'Target')
X_train2, X_test2, y_train2, y_test2 = split_data(df_SARCO, 'Target')
X_train3, X_test3, y_train3, y_test3 = split_data(df_Rabigh, 'Target')
X_train4, X_test4, y_train4, y_test4 = split_data(df_Bahri, 'Target')
X_train5, X_test5, y_train5, y_test5 = split_data(df_Aldrees, 'Target')
X_train1.head()
| Volume | EMA_9 | SMA_5 | SMA_10 | SMA_15 | SMA_20 | SMA_25 | SMA_30 | MACD | RSI | Stochastic | Open-Close | High-Low | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 8424 | 3790000 | 40.774786 | 44.850 | 38.670 | 45.090000 | 41.8670 | 41.7500 | 45.152000 | -2.486501 | 48.803456 | 25.520598 | 0.14 | 0.23 |
| 9645 | 2570000 | 59.698931 | 58.988 | 59.013 | 58.988667 | 59.2975 | 59.6008 | 59.751000 | -0.347814 | 48.291729 | 12.703849 | -0.19 | 0.23 |
| 10263 | 15450000 | 61.856118 | 75.372 | 61.934 | 57.569333 | 57.8500 | 57.7796 | 57.446333 | 3.569465 | 48.372071 | 19.749951 | 0.72 | 1.45 |
| 7553 | 25840000 | 30.043512 | 28.156 | 32.183 | 31.109333 | 31.1815 | 30.6312 | 30.462000 | -0.366022 | 49.312072 | 43.703129 | 1.73 | 1.82 |
| 7974 | 5530000 | 29.679506 | 29.534 | 29.995 | 30.113333 | 30.6495 | 31.2876 | 30.480000 | -0.869097 | 50.056896 | 50.632603 | -0.04 | 0.18 |
# Create a pipeline for the model, first scaling, then modeling.
pipe = make_pipeline(StandardScaler(), RF_grid)
pipe.fit(X_train1, y_train1)
pipe.fit(X_train2, y_train2)
pipe.fit(X_train3, y_train3)
pipe.fit(X_train4, y_train4)
pipe.fit(X_train5, y_train5)
print(f"model score: {pipe.score(X_test1, y_test1)}")
print(f"model score: {pipe.score(X_test2, y_test2)}")
print(f"model score: {pipe.score(X_test3, y_test3)}")
print(f"model score: {pipe.score(X_test4, y_test4)}")
print(f"model score: {pipe.score(X_test5, y_test5)}")
Fitting 10 folds for each of 72 candidates, totalling 720 fits Fitting 10 folds for each of 72 candidates, totalling 720 fits Fitting 10 folds for each of 72 candidates, totalling 720 fits Fitting 10 folds for each of 72 candidates, totalling 720 fits Fitting 10 folds for each of 72 candidates, totalling 720 fits model score: 0.5597014925373134 model score: 0.89738430583501 model score: 0.647887323943662 model score: 0.5774647887323944 model score: 0.7303822937625755
# to save the final model
import pickle
# create an iterator object with write permission - model.pkl
with open('model1_pkl', 'wb') as files:
pickle.dump(pipe, files)
# functions to process data, and prepare it for testing.
def prepare_data(df):
df['Date']=pd.to_datetime(df['Date'])
df.sort_values(by='Date', inplace=True)
df.reset_index(drop=True)
df['Tomorrow']=df['Price'].shift(-1)
df['Target']=(df['Tomorrow']> df['Price']).astype(int)
# change the type and remove % from the data.
df['Change %'] = df['Change %'].str.rstrip("%").astype(float)
# remove the string (K & M) to be able to change the type, then change the data type to int. ==> easier to deal with it
df["Vol."]=df["Vol."].replace({"K":"*1e3", "M":"*1e6"}, regex=True).map(pd.eval).astype(int)
# change symbol type to stirng.
df['Symbol']=df['Symbol'].astype(str)
# renaming columns to suitable names.
df.rename(columns={'Vol.':'Volume','Price':'Close'},inplace=True)
# drop unnecessary columns.
df = df.drop(['Sector','Industry'], axis=1)
featurecalculator(df)
#print(df.isnull().sum())
# Dropping rows with Na values
df=df.iloc[33:]
df.reset_index(drop=True,inplace=True)
df = df.fillna(method='ffill')
# add new features
df['Open-Close'] = df['Open'] - df['Close']
df['High-Low'] = df.High - df.Low
X = df.sort_values(by='Date', ascending=False).reset_index(drop=True).drop(['Tomorrow','Date','Name','Change %','Open','Close','High','Low','Symbol','Target'], axis=1) # drop the target
#print(X.head(2))
y = df['Target']
return df, X, y
# helper function
def featurecalculator(data):
data['EMA_9'] = data['Close'].ewm(9).mean() # exponential moving average of window 9
data['SMA_5'] = data['Close'].rolling(5).mean() # moving average of window 5
data['SMA_10'] = data['Close'].rolling(10).mean() # moving average of window 10
data['SMA_15'] = data['Close'].rolling(15).mean() # moving average of window 15
data['SMA_20'] = data['Close'].rolling(20).mean() # moving average of window 20
data['SMA_25'] = data['Close'].rolling(25).mean() # moving average of window 25
data['SMA_30'] = data['Close'].rolling(30).mean() # moving average of window 30
EMA_12 = pd.Series(data['Close'].ewm(span=12, min_periods=12).mean())
EMA_26 = pd.Series(data['Close'].ewm(span=26, min_periods=26).mean())
data['MACD'] = pd.Series(EMA_12 - EMA_26) # calculates Moving Average Convergence Divergence
data['RSI'] = RSIIndicator(data['Close']).rsi() # calculates Relative Strength Index
data['Stochastic']=StochasticOscillator(data['High'],data['Low'],data['Close']).stoch() # Calculates Stochastic Oscillator
pass
def test_data(model, df, X, y, no_of_preds):
y_pred = model.predict(X)
df_result = pd.DataFrame(y_pred, columns=['Predicted_Label'])
df_result['Target'] = y
df_result['Date'] = df['Date']
df_result.set_index('Date', inplace=True)
preds = df_result
acc_details = (df_result['Target'] == df_result['Predicted_Label']).value_counts()
acc = model.score(X, y)
print(f'Model Score: {round(acc*100, 2)}%')
print(f'Accuracy Details:\nFalse: {acc_details[0]}\nTrue: {acc_details[1]}\n')
print(f'Model Predications:\n {df_result[["Predicted_Label"]].tail(no_of_preds)}')
#return model_score, preds
# # load saved model
with open('model1_pkl' , 'rb') as f:
model = pickle.load(f)
df_test_Aramco = pd.read_csv('2222 Historical Data_test.csv')
df_test_Aramco, X, y = prepare_data(df_test_Aramco)
test_data(model, df_test_Aramco, X, y,5)
Model Score: 58.82%
Accuracy Details:
False: 28
True: 40
Model Predications:
Predicted_Label
Date
2022-08-29 0
2022-08-30 1
2022-08-31 0
2022-09-01 0
2022-09-04 1
df_test_SARCO = pd.read_csv('2030 Historical Data_test.csv')
df_test_SARCO.head()
df_test_SARCO, X, y = prepare_data(df_test_SARCO)
test_data(model, df_test_SARCO, X, y,6)
Model Score: 58.57%
Accuracy Details:
False: 29
True: 41
Model Predications:
Predicted_Label
Date
2022-08-30 0
2022-08-31 0
2022-09-01 0
2022-09-04 0
2022-09-05 0
2022-09-06 0
df_test_Rabigh = pd.read_csv('2380 Historical Data_test1.csv')
df_test_Rabigh, X, y = prepare_data(df_test_Rabigh)
test_data(model, df_test_Rabigh, X, y,1)
Model Score: 57.14%
Accuracy Details:
False: 30
True: 40
Model Predications:
Predicted_Label
Date
2022-09-06 1
df_test_Bahri = pd.read_csv('4030 Historical Data_test.csv')
df_test_Bahri, X, y = prepare_data(df_test_Bahri)
test_data(model, df_test_Bahri, X, y,5)
Model Score: 68.57%
Accuracy Details:
False: 22
True: 48
Model Predications:
Predicted_Label
Date
2022-08-31 1
2022-09-01 1
2022-09-04 1
2022-09-05 1
2022-09-06 1
df_test_Aldrees = pd.read_csv('4200 Historical Data_test.csv')
df_test_Aldrees, X, y = prepare_data(df_test_Aldrees)
test_data(model, df_test_Aldrees, X, y,5)
Model Score: 50.0%
Accuracy Details:
False: 35
True: 35
Model Predications:
Predicted_Label
Date
2022-08-31 0
2022-09-01 0
2022-09-04 0
2022-09-05 0
2022-09-06 0
def featurecalculator(data):
data['EMA_9'] = data['Close'].ewm(9).mean() # exponential moving average of window 9
data['SMA_5'] = data['Close'].rolling(5).mean() # moving average of window 5
data['SMA_10'] = data['Close'].rolling(10).mean() # moving average of window 10
data['SMA_15'] = data['Close'].rolling(15).mean() # moving average of window 15
data['SMA_20'] = data['Close'].rolling(20).mean() # moving average of window 20
data['SMA_25'] = data['Close'].rolling(25).mean() # moving average of window 25
data['SMA_30'] = data['Close'].rolling(30).mean() # moving average of window 30
EMA_12 = pd.Series(data['Close'].ewm(span=12, min_periods=12).mean())
EMA_26 = pd.Series(data['Close'].ewm(span=26, min_periods=26).mean())
data['MACD'] = pd.Series(EMA_12 - EMA_26) # calculates Moving Average Convergence Divergence
data['RSI'] = RSIIndicator(data['Close']).rsi() # calculates Relative Strength Index
data['Stochastic']=StochasticOscillator(data['High'],data['Low'],data['Close']).stoch() # Calculates Stochastic Oscillator
pass